Who wins in renewable energy? Evidence from Europe and the United States

Energy Research & Social Science 37 (2018) 65–73

Contents lists available at ScienceDirect

Energy Research & Social Science journal homepage: www.elsevier.com/locate/erss

Original research article

Who wins in renewable energy? Evidence from Europe and the United States a

Nina Kelsey , Jonas Meckling a b

b,⁎

MARK

Elliott School of International Affairs and Trachtenberg School of Public Policy, George Washington University, 1957 E St. NW, Washington, DC 20052, United States Dept. of Environmental Science, Policy, and Management, 130 Mulford Hall #3114, University of California, Berkeley, CA 94720, United States

A R T I C L E I N F O

A B S T R A C T

Keywords: Energy transition Renewable energy Solar Wind Political economy Electric utilities

The emerging transition to renewable energy, such as wind and solar photovoltaics, creates winners and losers in electricity markets. The political battle unfolds largely between incumbent electric utilities on the one hand and challenger firms such as independent power producers on the other. Here, we provide the first cross-national study of renewable energy ownership, based on an original dataset of fifty-nine jurisdictions in Europe and the United States. We find that independent power producers operating utility-scale generation dominate renewable energy capacity across electricity markets. Incumbent utilities and small producers of distributed generation hold substantially less capacity. Counter to expectations, this global trend is largely independent from two basic policy choices: the choice of support policy—feed-in tariffs versus renewable portfolio standards—and the choice of electricity market policy—liberalization versus regulation of power markets—only explain marginal effects on distributional outcomes. Rather, the resource potential of jurisdictions, relative technology prices, and the market effects of technological disruption likely account for the rise of medium-sized and large independent power producers as the dominant players in the transition to renewable energy. The transition to sustainable energy thus follows a substitution path, in which challenger firms prevail over incumbent utilities in renewable energy.

1. Introduction New renewable energy,1 i.e., wind and solar photovoltaics (PV), accounted for 17% of global renewable electricity generation in 2014, and is projected to grow to 42% by 2040 ([1], 412). The deployment of renewable energy contributed to the slowing growth of CO2 emissions in 2014 and 2015 [2]. While environmentally beneficial, the emerging transition to renewable energy creates, however, winners and losers in electricity markets around the globe. The political battle has been unfolding largely between incumbent electric utilities—which dominated power markets prior to the adoption of renewable energy policy—on the one hand, and challengers such as independent power producers (IPPs) and owners of small-scale distributed generation on the other.2 Who wins and loses in the rise of renewable energy technologies critically shape the political coalitions in favor or against the continued

⁎

transition toward sustainable energy [3–6]. The distribution of the benefits and costs of sustainable energy transitions affect in particular the durability of political support for such transitions [7,8]. This raises the question: Who wins in renewable energy, and why? What are the distributional dynamics of sustainable energy transitions? This article provides the first cross-national study on renewable energy ownership in 59 power markets in the EU (18) and the US (41), covering more than 95% of both wind and solar PV capacity in the two regions.3 We find that challenger firms—specifically IPPs operating utility-scale generation (USG)—dominate renewable energy capacity in the large majority of markets. Incumbent electric utilities, by contrast, hold only marginal shares in renewable energy capacity, which contrasts with their large majority shares in total power capacity (see Fig. 3). In short, IPPs with USG generation dominate renewable energy ownership in Europe and the United States, while incumbent utilities

Corresponding author. E-mail addresses: [email protected] (N. Kelsey), [email protected] (J. Meckling). We use “RE” to denote specifically wind and solar PV generation and capacity. We exclude other forms of generation, such as concentrating solar power and biomass, as these typically make up a small share of generation and comprehensive EU data for these technologies are not available. We also use “RE” only with reference to electricity generation, not transport fuels. 2 The literature defines incumbents and challengers in different ways. Here, we consider incumbents as those actors owning the large majority of generation capacity prior to the adoption of renewable energy policy (here, RPS or FIT). They thus held the greatest market power historically. As our data demonstrate, these were electric utilities; indeed, as we show in Fig. 3, utilities still retain the majority of conventional generation assets. All other actors are therefore by our definition challengers, even if they owned some conventional generation assets prior to the rise of renewable energy. 3 Asset ownership is a proxy for the distributional outcomes of the rise of RE technologies in electricity markets. 1

http://dx.doi.org/10.1016/j.erss.2017.08.003 Received 7 April 2017; Received in revised form 9 August 2017; Accepted 22 August 2017 2214-6296/ © 2017 Elsevier Ltd. All rights reserved.

Energy Research & Social Science 37 (2018) 65–73

N. Kelsey, J. Meckling

transform themselves [12]. We build on these notions of transitions, but note that here we conceptualize challengers to include firms that existed prior to the emergence of renewable energy but held minority shares in power capacity. This definition allows us to capture the overall trend of utilities losing out to IPPs in renewable energy markets. As we discuss below, IPPs are a broad set of actors, however, which is likely to result in a range of different substitution pathways. These include more corporate-driven and more citizen-driven paths to the substitution of incumbent utilities. Research on incumbent-challenger dynamics highlights a range of potential explanatory factors, including the industry setting, incumbent firm properties, and the nature of the challenge [13]. Here, we focus on the industry setting, in particular the institutional environment. Unlike other technological transitions, the emerging transition toward sustainable energy is driven primarily by government policy [14,15]. This raises the question whether policy choice shapes the extent to which incumbents and challengers partake in the emerging technological regime.

partake only marginally in it. Meanwhile, distributed generation, i.e., renewable energy assets owned by small producers, is growing, but in most markets it does not rival IPP-owned USG. We thus show that the transition toward renewable energy thus far appears for the most part to be following a “substitution” pathway in which challenger actors substitute for incumbent utilities in the new technology regime (cf. [9]). Our analysis examines policy-related as well as price and technology-related explanations of the rise of IPPs and USG in renewable energy. As regards policy, we explore whether renewable energy policy choice—feed-in tariffs (FIT) or renewable portfolio standards (RPS)—result in different ownership structures. We also examine the effect of electricity market policy—liberalization or regulation of power markets—on ownership structure. We find that both types of policy choices have only marginal direct effects on the market shares of incumbents and challengers in renewable energy. Instead, we find that resource endowments in wind, relative technology prices of wind and solar PV, and the dynamics of technological disruption are more likely to account for the rise of medium-sized and large IPPs as the dominant players in renewable energy. In outliers—markets where utilities or small producers do hold large shares in renewable energy capacity—the specific policy design, including the combination of several policy instruments, such as renewable portfolio standards with renewable energy certificates, is more likely to have shaped the distributional outcome than basic policy choice. In other words, policy helps explain outliers as opposed to the broad trend toward IPP ownership. Our findings have implications for policy. Moving beyond comparative cases, we observe a broad trend toward the substitution of incumbent utilities by challenger IPPs in the transition toward renewable energy in the EU and the US. This raises questions on how socially desirable different transition pathways are, in particular a more disruptive substitution pathway that leads to the decline of incumbent firms and the rise of new players versus a more incremental transformation pathway that results in incumbents adapting to the new technology. The implications are far-reaching, likely shaping market structure for decades to come. While our analysis suggests that the scope for policy to shape the pathway has limits, a more explicit debate on the desirability of different distributional outcomes in energy transitions is warranted. This article proceeds as follows. First, based on prior literature we develop expectations on the effect of policy choice on the ownership of renewable energy capacity. Second, we discuss our case selection and data collection. In a third step, we present our findings on the dependent variable, i.e., ownership structure in renewable energy capacity in Europe and the United States, and test our expectations. We also examine outlier cases. The conclusion summarizes the results and identifies the implications of our findings for the politics of sustainable energy transitions.

2.1. Policy choice and ownership distribution An extensive body of research has examined the relationship between policy and renewable energy. This includes the question of what drives the adoption of renewable energy policy [16,17]. Research has found political factors such as interest groups, political ideology, ruling party, and the policies of peer jurisdictions to play a role in renewable energy policy adoption [18–22]. Studies have also identified economic and resource-related drivers of government support for renewable energy, including market structure and resource endowments [20,21]. As renewable energy deployment has grown rapidly since the early 2000s, research has started to examine the effect of different types of policies on the level of deployment [23–25]. It also analyzed how different types of power market actors, such as investor-owned versus public utilities, respond to renewable energy policy [26]. This body of literature has only begun to consider the drivers of distributional outcomes in renewable energy transformations. We identify two main assumptions on the relationship between policy choice and why challengers—here, mostly IPPs—or incumbents—here, electric utilities—dominate in sustainable energy transitions. Those suggest that the choice of (1) renewable energy support policy and (2) electricity market policy are likely to shape which actors win and lose in renewable energy markets. First, the two most prominent support instruments for renewable energy are renewable portfolio standards and feed-in tariffs [27]. Renewable portfolio standards are thought to favor deployment of USG renewables by large producers—for reasons of greater economies of scale and their ability to manage the risk attached to investments under renewable portfolio standards [24,28,25,29]. Also, portfolio standards typically directly target utilities, although those utilities can opt to meet requirements by owning plants themselves or by buying electricity from IPPs. Feed-in tariffs, in contrast, are understood to provide in particular incentives for comparatively small producers such as households and small and medium-sized enterprises [30,31,23,12]. We would, therefore, expect quotas and renewable portfolio standards to favor utility-scale deployment, while feed-in tariffs favor higher levels of DG. Our second expectation relates to electricity market policy, i.e., whether a market is regulated or liberalized. The degree of liberalization of a market is understood to have a strong impact on market structure, with competition and monopolies at either end of the spectrum. Research has, for instance, shown that high concentration of market actors reduces the likelihood of renewable energy policy adoption [21]. Here, we extend this line of exploration to effects on ownership structure. The liberalization of electricity markets exposes incumbent utilities to competition from new entrants [32]. In fact, the historical evidence suggests that this shifts generation assets to IPPs [33]. We, therefore, expect that in regulated power markets incumbent

2. Sustainable energy transitions and distributional outcomes The literature on transitions sheds light on the dynamics of structural industrial and technological change. It highlights conflict between incumbent firms and challenger firms as a defining feature of transformational technological change within industries [10,11]. Depending on the relationship between incumbents and challengers, scholars identify different pathways of transitions [9]. A “substitution” pathway, for instance, suggests that new entrants to the market substitute incumbent players.4 Research suggests this is the case in the transformation of the German electricity sector. A “transformation” pathway, instead, unfolds when incumbent players adopt the new technology and 4 Our definition of substitution focuses on whether challengers come to dominate renewable energy ownership as the electricity industry shifts toward renewable energy, not on ownership of total generation capacity including legacy conventional capacity. Also, we examine only generation capacity, not shares in the retail market.

66

Energy Research & Social Science 37 (2018) 65–73

N. Kelsey, J. Meckling

Estonia, Ireland, Lithuania, Luxembourg, and Malta); for the most part these states were on the small end of the spectrum in terms of their capacity shares. Our final dataset consists of 41/50 US states and 18/28 EU states. Because our selection considered wind and solar separately, each standard (95% wind/95% solar) included some jurisdictions that would not have been included based on the other. Therefore our total coverage of each technology is in practice greater than 95%. This remains true even after the exclusion of the six jurisdictions with incomplete data, which account for roughly 2.3% and 0.3% of European wind and solar capacity, respectively.

utilities are more likely to dominate renewable energy markets, whereas in deregulated power markets IPPs are more likely to dominate renewable energy capacity. 2.2. Technology prices, resources, and ownership distribution Apart from policy choice, we expect that economic and resourcerelated factors could matter in shaping distributional outcomes. Renewable energy includes both wind and solar PV, each of which comes with different scale properties that lend themselves potentially to different types of ownership. For instance, wind is deployed largely as USG, whereas solar is deployed as both DG and USG. Depending on the relative costs of wind and solar PV technology, we would see different types of projects and, thus ownership types, prevail. In particular, all else being equal, if the cost of technology per unit generated for wind is lower than that cost of technology per unit generated for solar PV, we expect greater wind (and hence USG) deployment overall. We can assume a similar logic with regard to resource endowments. Resource endowments have been shown to affect the adoption of renewable energy policy: the greater resource endowment, the more likely a jurisdiction is to adopt renewables policy [20,21]. The argument could also hold for distributional outcomes. The relative endowment with wind potential and solar radiation potential could affect the relative deployment of the two technologies. Given that wind is almost exclusively deployed as USG, all else being equal, we would expect high wind potential to result in high USG deployment, and high solar potential to result in DG deployment.

3.2. Data collection Our dataset draws on (1) capacity data at the level of individual generation units, (2) data on policy adoption, and (3) data on resource potential. First, capacity data reports nameplate capacity of on-grid power plants that were operational in 2013. We aggregated capacity data for individual generation units to the plant level and considered any plant with a capacity greater than 1 MW utility-scale and with a capacity of 1 MW or less as DG. There is no universally agreed definition for the dividing line between utility-scale and distributed generation. A variety of authors have created useful overviews of different definitions for distributed generation in current use (see for instance [35,36]). Broadly, these definitions use different criteria to distinguish centralized power of the type owned (or contracted) and dispatched by utilities from smaller generation sources that are intended to serve local needs, including but not necessarily limited to net-metered power. Defining an upper bound for size is one option, although the bounds chosen differ fairly widely, from 1 MW to hundreds of megawatts. Other approaches focus on characteristics of the generation source in question, such as ownership (utility vs. other); the area of power distribution (local or long-distance); or the type of connection (to the distribution network or customer side of the meter rather than to the transmission network). We are particularly interested in the relative market shares of large producers that sell to the wholesale market and very small producers that generate power primarily for self-consumption. Our interest derives from the fact that there are major qualitative differences in the type of actor and business models involved in these two types of generation. Our category of “large producers” thus encompasses a broad range of large and medium-sized producers, all of which sell to the wholesale market. Our category of “small producers” is intended to encompass primarily residential and commercial producers that are definitely or likely generating power for self-consumption. We are not, for instance, interested in classing utility-owned assets used to balance load in the distribution system as small producers; and it is also the case that, until quite recently, even many renewable energy assets intended for wholesale production were relatively small and not necessarily owned by utilities or the largest IPPs. We are however limited by practical considerations: we do not have the data and resources necessary to consider and categorize each individual asset within our regions of interest on a case-by-case basis. This argues for a strategy that uses relatively clear dividing lines such as a numerical cut-off and characteristics readily derived from the data available. In practice, we use a combination of size (1 MW) and ownership type (utility/independent power producer/net metered/other distributed). (See Supplementary analysis for a more thorough discussion of this choice of cut-off.) It is of course also important to note that renewable energy ownership is only a proxy for distributional outcomes. Who reaps the profits in power generation may also determined by additional factors than actual ownership of generation capacity [25]. The sources for capacity data differ for the EU and the US. For the EU, we drew on datasets from the European Photovoltaic Industry

3. Methods In the following, we discuss case selection and data collection for our study of renewable energy ownership in Europe and the United States. 3.1. Case selection We selected the EU and the US as the two key regions for our study for three reasons. First, the goal was to test whether support policy choice and power market liberalization correlated with distributional outcomes in renewable energy markets. EU and US states provide variation across both variables. Second, our data set on the EU and the US accounts for more than 60% of total global renewable energy capacity in 2013 [34]. Finally, more comprehensive capacity data coded by ownership type is available in these two key deployment regions. Within the EU and the US, we selected country cases to include 95% of 2013 renewable energy generation capacity separately for both wind and solar PV in each of the two regions. For that purpose, we ranked countries by solar PV and wind shares in total generation capacity. We started with the market with the highest share and expanded toward lower shares of wind and solar PV until at least 95% of regional market coverage for each technology was achieved (see our Supplementary analysis for a more in-depth explanation of this inclusion process). The purpose of this selection was to exclude from our dataset cases with very low power market share in both solar and wind. Such states tend to distort comparative analyses of market share because their extremely small levels of deployment make them very vulnerable to noise from individual projects, a situation that does not shed light on our core analytical question. For instance, a state that shows 100% DG market share simply because there has been little or no meaningful renewable energy development overall, but there is a very small amount of existing grid-connected small solar PV, produces an extreme data point that is not representative of the bulk of the capacity deployment data. What the analyst should conclude in such a case is not that DG has been favored but rather that conditions are very poor for renewable energy generally. Following this case selection step, we also eliminated a small group of EU states where we lacked complete solar data (Cyprus, 67

Energy Research & Social Science 37 (2018) 65–73

N. Kelsey, J. Meckling

Third, to measure resource endowments at the country level, we used a variety of data, which we transformed into rank ordered data. Our use of rank-ordered data rather than absolute numbers is dictated by the fact that rank-ordered data was the best available data for solar potential in the EU. Hence, we transformed our data on wind in both regions, and on solar in the US, into ranked data to allow for parallel analysis. For solar in the EU, the country ranking by resource potential we used was developed by the European Commission [37]. Their ranking is based on the potential capacity factor, i.e., the technical potential of generation divided by the technical potential of capacity in a given state or country. For solar in the United States, we based our ranking on the state-specific capacity factor given in NREL analyses of solar technical potential [38].6 For wind in the EU, we calculate the potential capacity factor based on data from the European Environment Agency [39]; specifically, we rank countries based on EEA’s estimated potential generation capacity competitive as of 2020 divided by total land area. In wind in the US, we use NREL measurements of windy land area (windy defined as > =30% gross capacity factor for a turbine at 80 m) in a given state divided by total land area in the state [38]. Finally, we note that our total population is 41 US states and 18 EU countries. This constitutes most of the universe of currently existing useful cases with reliable data available. Our dataset is thus relatively small, which places natural limitations on the certainty of our findings and our ability to conduct more complex statistical analyses. This is particularly true given that we think it appropriate to treat the US and EU datasets separately (see Supplementary analysis for a discussion of this choice). As a result, although we provide statistical as well as descriptive analyses, we caution against over-interpreting the meaning of statistical significance per se in this context. In spite of this, as data begins to come in on the real-world distributional impact of the growth of renewable energy, we believe it is important to begin assessing what those impacts are. Policy decisions are being and must continue to be made. With appropriate conceptual caution, it can be useful to look more descriptively at the patterns of existing data to see if they generally support or conflict with our hypotheses. Hence, we employ descriptive statistics, some qualitative research, and simple statistical analysis to inductively identify the drivers of particular distributional outcomes across our cases [40].

Association (EPIA), the European Wind Energy Association (EWEA), and the proprietary Platts World Electric Power Plant Database (WEPP). For solar PV, data from EPIA provided total solar PV deployment by state and the percentage of that total that was USG/DG. WEPP data, which codes owner type by state, provided the basis for disaggregating the USG segment into utility, IPP, and other. The WEPP database provides representative samples rather than absolutely comprehensive data, but is the most comprehensive global power plant database available for the EU. For wind in the EU, EWEA reports utility-scale wind capacity only. We disaggregated the utility-scale data with the WEPP database, as in the case of solar. For DG wind, we faced data limitations. To our knowledge, there are very few national associations and statistics and no single entity that report DG wind for EU member states. We therefore resorted to a second-best method by conducting interviews with experts of wind industry associations to identify European member states in which DG wind, i.e., with a plant size of 1 MW or lower, was a meaningful share of wind generation. Denmark, Germany, and the United Kingdom were identified as such cases. For these countries, we drew on statistics of national industry associations or consulting firms to identify the size of the DG wind segment. For the remaining countries, we assumed zero capacity DG wind. United States capacity data for both wind and solar was drawn from US Energy Information Agency (EIA) databases. EIA provides data on utility-scale capacity, disaggregated by technology and producer, which we used to calculate shares of utility vs. IPP capacity. EIA also provides plant-level information on utility-scale capacity, which we used to further disaggregate IPP data by producer size. Utility-scale data covers grid-connected capacity of 1 MW or greater in size. EIA provides separate information on net-metered and distributed (non-utility-scale) generation capacity, broken down by technology and sector. This data set provides our data on grid-connected distributed generation in the United States. We do not include non-grid connected distributed capacity in our data, as comprehensive information on distributed nongrid connected capacity does not exist. Our analysis and conclusions throughout relate to grid-connected capacity specifically.5 Second, data on renewable energy policy adoption in the EU is based on the REN 21 Global Status Report. Data on power market liberalization for EU member countries was collected from a variety of sources, given the lack of a central data source. All sources are listed in the Supplementary dataset. Data on renewable energy support policy adoption in the US is based on data sets from the EIA on the adoption of renewable portfolio standards, feed-in tariffs, and feed-in-tariff-like schemes, supplemented in some cases with additional ad hoc research to confirm dates and specifics. Data on liberalization is similarly based on EIA reports on the status of restructuring across US states, supplemented with additional research to confirm dates and specifics such as critical legislation and implementation and suspension. We code a market as liberalized if both wholesale and retail markets have been liberalized. In the EU liberalization occurred in response to the EU’s Electricity Directive of 1996, though member states implemented it at varying speeds. In the US full liberalization, including retail choice, was left as a state-level decision, and only some states have adopted it. We note in passing that the reader may reasonably question whether wholesale-only integration is a better standard for this analysis than full liberalization (wholesale and retail). We address this in detail in our Supplementary analysis, but in short, we believe there are theoretical reasons why full liberalization is relevant to both generation and retail; moreover, while there is no relevant variation in the EU, testing more wholesale-specific definitions of liberalization in the US did not yield substantively different results.

4. Results: independent power producers win big, but why? This section presents our results in four steps. We first establish the dependent variable, i.e., ownership structure in renewable energy. We then show that policy-related expectations explain only marginal effects in the distribution of ownership, but that price and technology-related factors appear to hold greater explanatory power. We conclude by discussing outliers, cases where incumbent utilities win big or DG dominates. 4.1. Independent power producers in renewable energy Utility-scale renewable energy plants owned by IPPs are the dominant category of grid-connected renewable energy assets based on generation capacity. This suggests that medium-sized and large new entrants are the major “winners” in the rise of renewable energy generation capacity. IPPs include a large number of private power producers that either sell power to the wholesale market or enter long-term 6 We opt here for fairly general measurements of potential – later in our analysis we ask whether the resource potential of a jurisdiction offers an explanation for wind and solar deployment, with the understanding that wind and solar are strongly linked to USG and DG respectively. As such, we have chosen fundamental measures such as average irradiation and percent windy land area. These measures are also similar to the approach used to create the solar potential ranking for EU countries, where only the rank ordering is available to us. However, please see our Supplementary analysis for a discussion of other possible approaches, some of which offer potential avenues for further research.

5 Please see our Supplementary analysis for further discussion of off-grid capacity and its implications for our analysis; in brief, the available data suggests that off-grid capacity is not a major part of the story in either solar or wind.

68

Energy Research & Social Science 37 (2018) 65–73

N. Kelsey, J. Meckling

contractual agreements with electric utilities, such as cooperatives and commercial power producers. Incumbent utilities, and individual household and small enterprise auto-producers participate in renewable energy to a lesser extent, although differences exist between wind and solar PV. USG dominates in 76% of our 59 country/state cases, while DG dominates in only 15% (the rest are roughly equal). In the EU, on average across 18 cases 71% of renewable energy capacity is utilityscale; in the US, on average across our 41 cases 83% is utility-scale. We find that IPPs tend to be winners relative to both competitors in USG (utilities) and competitors from the world of distributed generation. While both utilities and IPPs operate USG plants, IPPs are by far the dominant players within USG renewable energy: in 90% of our cases, IPPs hold substantially more USG renewable energy capacity than utilities (see Fig. 2). This contrasts with utilities’ majority shares in overall USG power generation assets. And IPP USG share by itself is also typically larger than total DG: small producers, i.e., those operating DG for self-consumption (capacity of 1 MW or less), also account for substantially less capacity than IPPs alone in 73% of all cases. These outcomes hold across the EU and the US, although the EU cases are on average somewhat more favorable to DG. The high share of IPPs in renewable energy markets contrasts with their lower shares in non-renewable power generation capacity including fossil fuels. Utilities on average hold well over half of all nonwind, non-solar-PV USG capacity in the US, and almost three quarters in the EU; but only about 16% and 14% of USG renewable energy capacity in the EU and the US respectively (see Fig. 3). While IPPs are the primary winners in renewable power across the EU and the US, they constitute a broad group of different producers, including specialized renewable energy wholesale producers and community-owned cooperatives [41]. In fact, the median capacity per IPP varies between the EU and the US. In the US, the median of the state averages of renewable energy capacity per IPP is 67 MW, whereas the median of country averages of renewable energy capacity per IPP is 38 MW in the EU. IPPs in the US, thus, tend to be larger than those in the EU.

Fig. 2. IPP and utility renewable energy shares of total USG (all USG generation types). Legend: Bubble size indicates country/state percent of total RE USG (wind + solar PV) capacity within dataset.

portfolio standards than under feed-in tariffs in both the EU and the US (with cases with both policies falling in the middle; see Fig. 4). The difference between categories is non-significant in the EU (Kruskal-Wallis, H = 1.272, 2 d.f., p = 0.53). And while it is initially significant for the US (Kruskal-Wallis, H = 7.233, 2 d.f., p = 0.03) further testing with Bonferroni correction shows that is driven entirely by differences between the “Neither” category and others; the difference between “RPS Only” and “FIT + RPS” is non-significant (MannWhitney U = 64, nRPS = 24, nBoth = 6, p = 0.68 two-tailed).7 Hence, any actual impact of instrument choice is uncertain and marginal. We do believe that the consistent trends across categories in both the US and EU are suggestive that renewable portfolio schemes likely do support USG slightly more than feed-in tariff policies. But given that the differences are small and non-significant, we conclude that the observed trends are consonant with conventional wisdom, but differ in that any effect is small at most for renewable energy. Second, the literature suggests that regulated markets would lead to a greater role for utilities relative to IPPs, while liberalized markets would lead to the opposite. Here again, evidence for this expected effect is weak. EU states provide no evidence for the hypothesis that utilities are less likely to win in states liberalized at the time of renewable energy policy imposition: In fact, in markets liberalized at the time of the adoption of renewable energy policy, utilities own on average 19% capacity as opposed to 12% in markets not liberalized at the time of renewable energy policy adoption, though the difference is non-significant (Mann-Whitney U = 23, nLib = 10, nNotLib = 8, p = 0.13 twotailed). Those states with non-liberalized power markets at the time of renewable energy policy adoption are mostly early adopters of feed-in tariffs, such as Germany (1990), Spain (1994), and Italy (1991). This counter-intuitive result may be because renewable energy policy in those cases created de facto liberalization in renewable energy generation by removing market entry barriers for renewable energy generation. In contrast, full liberalization at the time of renewables policy adoption is weakly associated with lower utility share in the US. In liberalized markets, utilities average 6% of renewable energy USG, while they average 19% in unliberalized markets, a difference that is not significant at the 0.05 level, although we note that it comes

4.2. Limited distributional effects of policy choice We examine whether policy choice affects renewable energy ownership. We only find marginal effects of renewable energy policy and of electricity market liberalization on ownership structure. First, existing literature suggests that renewable portfolio standards would favor USG, while feed-in tariffs would lead to more DG. As Fig. 1 above suggests, in more than three-quarters of all of our cases renewable energy USG is substantially greater than renewable energy DG. Within this overall pattern, we do observe that the share of USG is slightly greater under

7 This analysis produces similar results with the EIA “distributed” class assigned to the USG category rather than the DG category as discussed above in the section on our definitions of USG and DG. In addition to utility-scale and net-metered generation capacity, the EIA’s data contains a somewhat debatable class of non-utility scale generation assets they label “distributed.” These can arguably be assigned either to the USG or DG categories, since they are smaller but not net-metered. Carrying out this analysis with either assignment produces similar results.

Fig. 1. USG vs. DG renewable energy market share by jurisdiction. Legend: Bubble size indicates country/state percent of total RE (wind + solar PV) capacity within dataset.

69

Energy Research & Social Science 37 (2018) 65–73

N. Kelsey, J. Meckling

Fig. 3. IPP and utility shares in non-renewables versus renewables.

Most utility-scale generation is wind, while most distributed generation is solar. And wind is deployed more widely than solar. The levelized cost of wind has historically been substantially lower than that of solar, achieving price parity with conventional fossil fuels in some locations [42].8 Hence it makes simple economic sense to deploy wind in windy areas, regardless of policy. And indeed, the availability of wind resources correlates well with actual wind deployment; the greater the technical wind resource potential of a jurisdiction, the more wind has been deployed (see Fig. 5) and hence (since wind is largely USG) the more prevalent USG is. In other words, high deployment of wind explains the prevalence of USG, and the availability of wind explains levels of wind deployment. This resonates with Lyon and Yin’s [20] finding that renewable energy potential is one of the key drivers of the adoption of renewable portfolio standards. The correlation between wind potential rank and wind actual deployment rank is strongly significant in both the US (rs = 0.779, p < 0.001) and EU (rs = 0.660, p = 0.003). Given our ranked data, we perform Spearman rank correlation tests on this data and do not report regression coefficients.9 The picture is quite literally different in the case of solar (which is primarily distributed generation) deployment. There is clearly no correlation between solar potential and actual deployment in the US subset (rs = 0.049, p = 0.76); and no significant relationship in the EU (rs = 0.243, p = 0.33) (Fig. 6). The greater variation of outcomes in solar deployment relative to resource potential suggests that solar deployment is more dependent on factors other than geography and economics, including policy drivers. Solar has been economically less competitive than wind, meaning that even where solar resources are

Fig. 4. Average USG shares of renewable energy capacity by policy category.

substantially closer than most of the results we report here, and given the small size of our data set might still be seen as suggestive depending on the reader’s approach to the use of p-values (Mann-Whitney U = 74.5, nLib = 16, nNotLib = 15, p = 0.067 two-tailed). The five states with utility shares of renewable energy that are at least equal to IPP shares are all unliberalized. Regardless of whether we consider this suggestive of a real relationship, however, it is again at best marginal relative to the general dominance of IPP ownership. On average, in both regions, IPPs are the “winners” in renewable energy either way. 4.3. Relative prices, resource endowments, and technological disruption

8 Of course, while relative costs of wind and solar are separate drivers from the support policy and liberalization choices we examine here, they are not necessarily entirely divorced from policy as a whole; for instance, policy choices around research funding could impact this cost disparity. 9 Since the Spearman rank correlation deals with ranked data points, rather than continuous data, the potential meaning of the regression coefficient is ambiguous and potentially misleading; as such, we do not report an equation and coefficient for our trendlines as we would with a standard linear regression.

While the choice of support instrument—RPS vs. FIT—may account for marginal effects in renewable energy asset ownership, it does not explain the major outcome of USG prevalence. Relative technology prices and resource endowments offer a more convincing explanation for why USG has emerged as the dominant type of renewable energy plant. 70

Energy Research & Social Science 37 (2018) 65–73

N. Kelsey, J. Meckling

Fig. 5. Wind capacity deployment, rank based on technical potential versus actual.

Fig. 7. Average solar USG shares by policy category.

the exception of offshore wind—have smaller average capacities than conventional power plants. This difference favors new entrants. New and smaller entrants could more easily fit their business models around this new type of generation, whereas utilities struggled fitting renewable energy technologies into their portfolio [43–45]. Some evidence also suggests that incumbent utilities underestimated the growth of renewable energy, leading them to not enter the segment aggressively. For instance, in Germany incumbent utilities underestimated the effects of the feed-in tariff, but were also distracted by the political battle over nuclear phase-out [46,47]. These initial insights suggest that factors related to the particular challenge of a transition toward renewable energy and the attributes of incumbent actors may account for the low level of participation of incumbent utilities in renewable energy (cf. [13]). This resonates with past experience of the emergence of new power technology. IPPs similarly dominated ownership of smaller, modular combined cycle natural gas plants when they emerged as competitors to larger-scale conventional power plants [48]. The challenge of technology transition is not necessarily entirely distinct from policy choices. Different policy regimes could potentially moderate or mediate the impacts of technology disruption on utilities, but the similar outcome in the prior gas case is suggestive that there really is a structural aspect here not dependent on policy. We cannot, however, draw overall conclusions across the entire sample of power markets in this study. This calls for future research, as we discuss in the conclusion. Renewable energy markets are also a dynamic field, and it remains to be seen if and how utilities adapt to the challenge.

Fig. 6. Solar PV capacity deployment, technical potential versus actual.

strong, solar is unlikely to be deployed without additional drivers. This means that solar is generally not deployed as much as wind, but when it is deployed, it is not deployed in a way that is clearly dictated by resource availability. A comparison of power markets with renewable energy policy and without renewable energy policy in the US (all EU power markets have renewable energy policy) also supports the idea that relative prices more than policy drive USG deployment. Renewable energy capacity amounts to 8% average market share across states with support policy versus a relatively similar 6% across states without support policy—a non-significant difference. Resource-rich states deploy wind with or without policy incentives. But solar PV taken alone differs. States with renewables policy have a 1.5% average solar market share compared to 0.1% across states without, a strongly significant difference. However, solar still does not display a clear relationship to renewable portfolio standard/feed-in tariff choice (See Fig. 7). Solar market share does show visible variation across policy categories, but not in a consistent or statistically significant manner for either the EU (Kruskal-Wallis, H = 1.863, 2 d.f., p = 0.39) or US (Kruskal-Wallis, H = 0.533, 2 d.f., p = 0.77). We conclude that policy in wind/USG-heavy cases is not a primary driver of the distribution of ownership in comparison to the effects of relative prices and resource endowments. Policy choices may matter in solar/DG-heavy cases, but not in a way clearly predicted by expectations about basic instrument choice that can be extrapolated from the literature. Turning to ownership within USG, if liberalization does not appear to be a strong driver, what explains the dominance of IPPs relative to utilities? A growing body of empirical work suggests that the nature of renewable energy technologies combined with an underestimation of the strength of the growth of renewable energy by utilities led to the distribution of ownership we observe. Wind and solar PV plants—with

4.4. Explaining outliers: policy design and distributional effects We identify two sets of outliers: markets with either a prevalence of DG, i.e. small producers prevail, or of utility renewable energy ownership. First, in a small set of cases, DG is equal to or larger than USG. In the EU, Denmark, Germany, and Italy have high shares of DG. These states were all first movers in RE policy adoption in the early 1990s and promoted decentralized energy ownership [49,50]. The particular design of their feed-in tariffs was especially favorable to DG, notably solar PV [31,51]. Policy design appears to play a similar role in the US. A cluster of states in the Northeast (especially CT, DE, MA, NJ, and RI) is particularly solar/DG heavy despite being fairly solar resource-poor. Feed-in tariffs do not seem to be a key driver: of these states, only Rhode Island has a feed-in tariff in place (along with a renewable portfolio standard). However, the existence of solar renewable energy credit (SREC) programs in several states in this area means utilities can use DG to meet renewable portfolio requirements, creating synergy between renewable portfolio standards and DG—a dynamic likely intensified by the lack of available land area for USG renewable energy in the densely settled northeast. 71

Energy Research & Social Science 37 (2018) 65–73

N. Kelsey, J. Meckling

strongly determine this distributional outcome. Feed-in tariff cases have only a marginally higher share of DG, i.e. capacity owned by small producers. This runs counter to assumptions that feed-in tariffs incentivize predominantly small producers to adopt renewable energy. Similarly, renewable portfolio standard cases are not strongly associated with higher shares of renewable energy owned by utilities. Rather than policy choice, economic and geographic factors appear to account for the rise of USG renewable energy and the fact large producers dominate renewable energy ownership. Supported by policy incentives, investment follows the lower cost of wind and resource potential by-and-large. However, in the small set of cases where DG has been particularly successful, the specific policy design and the combination of several policy instruments—such as renewable portfolio standards, SRECs, and net-metering incentives—appear likely to have shaped distributional outcomes. Policy choice and design thus appear most likely to matter in particular for the share of DG, i.e., by-and-large small solar producers, in renewable energy markets. Research on winners and losers in energy transitions is in its infancy. We identify in particular three areas for future research. First, while this paper shows that IPPs owning USG are the predominant actors in renewable energy, the category of IPPs covers a broad range of actors, including project developers, cooperatives, medium-sized and large firms. We believe it is useful to consider IPPs generally as challenger firms relative to entrenched utilities, but this does not mean that all challengers are created equal. A transition in which utilities are supplanted mostly by large, consolidated corporate merchant generators is from a transition in which utilities are supplanted mostly by smaller-scale community ownership, aggregators of distributed generation, or some other mix or variation. Case studies have examined different types of challenger-led transitions. Yet large-n analyses that disaggregate IPPs into different actor types would further advance our systematic understanding of challenger-led transitions. Second, while we observe a trend toward IPPs owning the majority of renewable energy assets, we also identify outliers where utilities own the majority of capacity. In-depth case studies of such transformation cases would offer critical insights into the preconditions of more transformative pathways, including company-specific factors as discussed above. Third, our analysis of outliers suggests that specific policy mixes may have an effect on ownership distribution. This speaks to recent debates on the role of policy mixes in energy transitions [56,57]. Linking specific combinations of policy instruments to distributional outcomes would advance our understanding of the potential for policy to influence who wins and loses in the rise of renewable energy. This would also call for indicators for distributional outcomes beyond ownership. Our findings have implications for policy. We observe a broad trend toward the substitution of incumbent utilities by challenger IPPs in the transition toward renewable energy in the EU and the US. The downfall of German utilities such as EON and RWE may just be the early signs of a much broader shift in power markets. Between 2008 and 2013, the top 20 European utilities lost more than half of their total stock market valuation, though only part of that decline can be attributed to the rise of renewable energy [58]. This raises questions on how socially desirable different transition pathways are, in particular a more disruptive substitution pathway versus a more incremental transformation pathway. This speaks to broader debates on the distributional dynamics of policy responses to climate change [59]. The implications of different ownership distributions are far-reaching, shaping market structures for decades to come. This warrants a more explicit debate on the desirability of different distributional outcomes in sustainable energy transitions. Our analysis suggests that policy leverage to shape the distribution of ownership may be asymmetrical. The trend toward IPPs as winners in terms of ownership is independent from basic policy choice. Policy design is, however, likely to be able to influence whether there is also a substantial constituency of smaller distributed self-producers and solar generation ownership.

Second, in some cases utility ownership within USG remains high relative to IPPs. In the EU, utilities in Denmark have a relatively high share in USG renewable energy capacity. Anecdotal evidence suggests that corporatist negotiations between the government and electric utilities led to an initial head start for utilities in renewable energy. In Denmark, the government and electric utilities negotiated an agreement to develop five offshore wind parks, the first of which went operational in 2002 [52]. In the Netherlands, the government also developed policy that was favorable to electric utilities [53–55]. Again, specific policy design—rather than basic policy choice—offers an explanation. In the US, no clear explanation emerges that applies to all highutility-ownership outliers, but a similar “head-start” effect may explain two cases. Iowa and Wisconsin are both cases in which some form of substantive, utility-focused renewable energy policy predates early wholesale market opening and the general rise of IPPs in the US. Iowa adopted a renewable portfolio standard early on, while Wisconsin prioritized renewable asset development by the public utility. These early efforts may have meant those utilities were better adapted to renewable energy and suffered less from the incumbent disadvantages discussed above. Utilities also own a substantial share of renewable energy capacity in Vermont, Washington, and Wyoming. We note that their utilityowned renewable energy capacity is largely in the hands of just three utility companies across these three states. Our data show that all of the utility-owned renewable capacity in Vermont comes from two wind farms owned by one utility.10 Almost all the utility-owned renewable capacity in Wyoming belongs to one company. Utility-owned renewable capacity in Washington is largely split between two utilities, one of which is the same company that controls the utility-owned renewable capacity in Wyoming.11 The high concentration of renewable energy ownership in these jurisdictions suggests the potential for companyspecific explanations, which warrants research into the factors affecting decision-making at the individual utility level in seeking explanations for outliers with high utility ownership shares. 5. Conclusion and implications Renewable energy is expanding rapidly, transforming the technical, economic, and political dimensions of power markets. As in all major industrial transformations, this creates winners and losers. Our analysis provides the most comprehensive coverage of ownership distribution to date. IPPs owning USG renewable energy are the overall “winners” in a large majority of renewable energy markets in the EU and the US. This is a uniform trend, though there are outliers. The findings support previous research that has suggested that political support for renewable energy policy comes from challengers in power markets [3,20,47]. Our data provides broad cross-country evidence that challenger firms are major beneficiaries. We thus observe by-and-large a substitution pathway in the shift toward renewable sources of electricity in Europe and the United States. Our data suggest that medium-sized and large (USG) IPPs are the primary beneficiaries, rather than small producers of DG. Renewable policy choice and power sector liberalization do not 10 Green Mountain Power Corp. Although there are other public power entities in Vermont – EIA data lists 9 entities in its electric utility category – Green Mountain Power is the largest in terms of customers and generation owned; the others are largely local coops and municipal power entities. 11 PacifiCorp, a company that owns generation capacity across several western states. In Wyoming, PacifiCorp is the largest utility owner of generation, although there are several smaller owners and one (Basin Electric Power Coop) owns about half the capacity that PacifiCorp does. In Washington, PacifiCorp is only one among a number of utilites with large total generation ownership (EIA has six entities in its utility category in Washington state that own more), although only one of these (Puget Sound Energy Inc.) owns more wind generation capacity specifically.

72

Energy Research & Social Science 37 (2018) 65–73

N. Kelsey, J. Meckling

support instruments, Energy Policy 45 (2012) 635–644. [26] M.A. Delmas, M.J. Montes-Sancho, U.S. state policies for renewable energy: context and effectiveness, Energy Policy 39 (5) (2011) 2273–2288. [27] REN21, Renewables 2015: Global Status Report Paris, Renewable Energy Policy Network for the 21 st Century, (2015). [28] J. Szarka, Climate challenges, ecological modernization, and technological forcing: policy lessons from a comparative US-EU analysis, Global Environ. Polit. 12 (2) (2012) 87–109. [29] J. Meckling, S. Jenner, Varieties of market-based policy: instrument choice in climate policy, Environ. Polit. 25 (5) (2016) 853–874. [30] N.I. Meyer, European schemes for promoting renewables in liberalised markets, Energy Policy 31 (7) (2003) 665–676. [31] I.H. Rowlands, Envisaging feed-in tariffs for solar photovoltaic electricity: European lessons for Canada, Renew. Sustain. Energy Rev. 9 (1) (2005) 51–68. [32] M.K. Heiman, B.D. Solomon, Power to the people: electric utility restructuring and the commitment to renewable energy, Ann. Assoc. Am. Geogr. 94 (1) (2004) 94–116. [33] S. Borenstein, J. Bushnell, The US electricity industry after 20 years of restructuring, Ann. Rev. Econ. 7 (1) (2015) 437–463. [34] IEA, World Energy Outlook, International Energy Agency., Paris, 2015. [35] T. Ackermann, G. Andersson, L. Söder, Distributed generation: a definition, Electr. Power Syst. Res. 57 (2001) 195–204. [36] G. Pepermans, J. Driesen, D. Haeseldonckx, R. Belmans, W. D’haeseleer, Distributed generation: definition, benefits and issues, Energy Policy 33 (2005) 787–798. [37] M. Šúri, T.A. Huld, E.D. Dunlop, H.A. Ossenbrink, Potential of solar electricity generation in the European Union member states and candidate countries, Sol. Energy 81 (10) (2007) 1295–1305. [38] A. Lopez, B. Roberts, D. Heimiller, N. Blair, U.S. Renewable Energy Technical Potentials: A GIS-Based Analysis, National Renewable Energy Laboratory, Golden, CO, 2012. [39] EEA, Europe's Onshore and Offshore Wind Energy Potential, European Environment Agency, Copenhagen, 2009. [40] K.M. Eisenhardt, M.E. Graebner, Theory building from cases: opportunities and challenges, Acad. Manage. J. 50 (1) (2007) 25–32. [41] M. Oteman, M. Wiering, J.-K. Helderman, The institutional space of community initiatives for renewable energy, Sustain. Soc. 4 (11) (2014) 1–17. [42] P.L. Joskow, Comparing the costs of intermittent and dispatchable electricity generation technologies, Am. Econ. Rev.: Papers Proc. 101 (3) (2011) 238–241. [43] K. Hockerts, R. Wüstenhagen, Greening Goliaths versus emerging Davids — theorizing about the role of incumbents and new entrants in sustainable entrepreneurship, J. Bus. Venturing 25 (5) (2010) 481–492. [44] K. Groot, European Power Utilities Under Pressure? CIEP Paper 3 Clingendael International Energy Programme, The Hague, 2013. [45] E. Graffy, S. Kihm, Does disruptive competition mean a death spiral for electric utilities, Energy Law J. 35 (1) (2014) 1–44. [46] M. Richter, German utilities and distributed PV: how to overcome barriers to business model innovation, Renew. Energy 55 (2013) 456–466. [47] G. Kungl, Stewards or sticklers for change? Incumbent energy providers and the politics of the German energy transition, Energy Res. Soc. Sci. 8 (2015) 13–23. [48] R.F. Hirsh, Power Loss: The Origins of Deregulation and Restructuring in the American Electric Utility System, MIT Press, Cambridge, MA, 1999. [49] M. Mendonça, S. Lacey, F. Hvelplund, Stability, participation and transparency in renewable energy policy: lessons from Denmark and the United States, Policy Soc. 27 (4) (2009) 379–398. [50] P.O. Eikeland, T.H.J. Inderberg, Energy system transformation and long-term interest constellations in Denmark: can agency beat structure, Energy Res. Soc. Sci. 11 (2016) 164–173. [51] A. Klein, E. Merkel, B. Pfluger, A. Held, Evaluation of Different Feed-In Tariff Design Options, Fraunhofer ISI, Karlsruhe, Germany, 2010. [52] N.I. Meyer, Renewable energy policy in Denmark, Energy Sustain. Dev. VIII (1) (2004) 25–35. [53] M. Wolsink, Dutch wind power policy, Energy Policy 24 (12) (1996) 1079–1088. [54] S. Agterbosch, W. Vermeulen, P. Glasbergen, Implementation of wind energy in the Netherlands: the importance of the social-institutional setting, Energy Policy 32 (18) (2004) 2049–2066. [55] S. Breukers, M. Wolsink, Wind power implementation in changing institutional landscapes: an international comparison, Energy Policy 35 (5) (2007) 2737–2750. [56] P. Kivimaa, F. Kern, Creative destruction or mere niche support? Innovation policy mixes for sustainability transitions, Res. Policy 45 (1) (2016) 205–217. [57] K.S. Rogge, K. Reichardt, Policy mixes for sustainability transitions: an extended concept and framework for analysis, Res. Policy 45 (8) (2016) 1620–1635. [58] The Economist, How to Lose Half a Trillion Euros: Europe's Electricity Providers Face an Existential Threat, The Economist, 2013 (October 12). [59] S. Klinsky, T. Roberts, S. Huq, C. Okereke, P. Newell, P. Dauvergne, K. O’Brien, H. Schroeder, P. Tschakert, J. Clapp, M. Keck, F. Biermann, D. Liverman, J. Gupta, A. Rahman, D. Messner, D. Pellow, S. Bauer, Why equity is fundamental in climate change policy research, Global Environ. Change 44 (2016) 170–173.

Acknowledgments Authors contributed equally. We are thankful for comments from Eric Biber, Sara Chatfield, Steffen Jenner, Corinna Klessmann, Matto Mildenberger and John Zysman and for research assistance by Jane Flegal, Avi Ranger, Lauren Murphy, and Sasan Saadat. We also gratefully acknowledge funding from the ClimateWorks Foundation. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.erss.2017.08.003. References [1] IEA, World Energy Outlook 2016, International Energy Agency, Paris, 2016. [2] R.B. Jackson, J.G. Canadell, C. Le Quéré, R.M. Andrew, J.I. Korsbakken, G.P. Peters, N. Nakicenovic, Reaching peak emissions, Nat. Clim. Change 6 (2015) 7–10. [3] F.N. Laird, C. Stefes, The diverging paths of German and United States policies for renewable energy: sources of difference, Energy Policy 37 (7) (2009) 2619–2629. [4] J. Meckling, N. Kelsey, E. Biber, J. Zysman, Winning coalitions for climate policy: green industrial policy builds support for carbon regulation, Science 249 (6253) (2015) 1170–1171. [5] M. Lockwood, C. Kuzemko, C. Mitchell, R. Hoggett, Historical institutionalism and the politics of sustainable energy transitions: a research agenda, Environ. Plann. C: Gov. Policy (2016) 1–22. [6] E. Schmid, B. Knopf, A. Pechan, Putting an energy system transformation into practice: the case of the German Energiewende, Energy Res. Soc. Sci. 11 (2016) (2016) 263–275. [7] M. Lockwood, The political sustainability of climate policy: the Case of the UK Climate Change Act, Global Environ. Change 23 (2013) 1339–1348. [8] E. Biber, N. Kelsey, J. Meckling, The political economy of decarbonization: a research agenda, Brooklyn Law Rev. 82 (2) (2017) 605–643. [9] F.W. Geels, J. Schot, Typology of sociotechnical transition pathways, Res. Policy 36 (3) (2007) 399–417. [10] A. Bergek, C. Berggren, T. Magnusson, Technological discontinuities and the challenge for incumbent firms: destruction, disruption or creative accumulation, Res. Policy 42 (6–7) (2013) 1210–1224. [11] Andy Stirling, Transforming power: social science and the politics of energy choices, Energy Res. Soc. Sci. 1 (2014) 83–95. [12] F.W. Geels, F. Kern, G. Fuchs, N. Hinderer, G. Kungl, J. Mylan, M. Neukirch, S. Wassermann, The enactment of socio-technical transition pathways: a reformulated typology and a comparative multi-level analysis of the German and UK low-carbon electricity transitions (1990–2014), Res. Policy 45 (4) (2016) 896–913. [13] S. Ansari, P. Krop, Incumbent performance in the face of a radical innovation: towards a framework for incumbent challenger dynamics, Res. Policy 41 (8) (2012) 1357–1374. [14] F. Kern, K.S. Rogge, The pace of governed energy transitions: agency, international dynamics and the global Paris agreement accelerating decarbonisation processes? Energy Res. Soc. Sci. 22 (2016) 13–17. [15] J. Meckling, The developmental state in global regulation: economic change and climate policy, Eur. J. Int. Relat. (2017), http://dx.doi.org/10.1177/ 1354066117700966. [16] B.G. Rabe, Second generation climate policies in the American States. proliferation, diffusion, and regionalization, Issues Gov. Stud. 6 (August) (2006) (2006). [17] K.S. Gallagher, Why & how governments support renewable energy, Daedalus 142 (1) (2013) 59–77. [18] D.C. Matisoff, The adoption of state climate change policies and renewable portfolio standards, Rev. Policy Res. 25 (6) (2006) 527–546. [19] J. Chandler, Trendy solutions.. Why do states adopt sustainable energy portfolio standards? Energy Policy 37 (8) (2009) 3274–3281. [20] T. Lyon, H. Yin, Why do states adopt renewable portfolio standards? An empirical investigation, Energy J. 31 (3) (2010) 131–155. [21] S. Jenner, G. Chan, R. Frankenberger, M. Gabel, What drives states to support renewable energy? Energy J. 33 (2) (2012) 1–12. [22] H. Yi, R.C. Feiock, Renewable energy politics: policy typologies, policy tools, and state deployment of renewables, Policy Stud. J. 42 (3) (2014) 391–415. [23] C. Mitchell, D. Bauknecht, P.M. Connor, Effectiveness through risk reduction: a comparison of the renewable obligation in England and Wales and the feed-in system in Germany, Energy Policy 34 (3) (2006) 297–305. [24] R. Green, A. Yatchew, Support schemes for renewable energy: an economic analysis, Econ. Energy Environ. Policy 1 (2) (2012) 83–98. [25] A. Verbruggen, V. Lauber, Assessing the performance of renewable electricity

73