Renewable energy auctions When are they (cost-)effective?

Article Renewable energy auctions When are they (cost-)effective? Energy & Environment 2017, Vol. 28(1 2) 145 165! The Author(s) 2017 Reprints and permissions: sagepub.co.uk/journalspermissions.nav DOI: 10.1177/0958305X16688811 journals.sagepub.com/home/eae Malte Gephart, Corinna Klessmann and Fabian Wigand Abstract European Member States increasingly use tenders combined with competitive bidding to allocate renewable electricity support payments to renewable electricity market actors. This article contributes to the European policy debate by exploring which principal design features and context factors increase or reduce the effectiveness and cost-effectiveness of renewable electricity auctions. Volume control is among the key aims of implementing an auction, but potentially low project realisation can harm reaching the targeted volume. Qualification requirements and/or penalties are useful auction design elements to increase the implementation rates of selected projects. However, there are risks associated with these measures, which in turn increase prices. Auctions also aim at increasing (static) cost-effectiveness of renewable electricity support, which may be influenced by three main factors: first, the level of competition in the auction; second, the mitigation of speculative over- or under-bidding; and third, the level of allocation and delivery risks borne by the bidders. The article explores the fundamental trade-off in auctions between encouraging high project implementation rates (to ensure volume control) and minimising the bidder s risk (that may result in higher bidding prices). Based on theoretical insights and supported by empirical renewable electricity auction examples, it identifies factors that can influence the success of an auction but also shows there is no exact blueprint for a good auction design. Keywords Auction, tender, competitive bidding, renewable electricity, support scheme Introduction In Europe, tenders combined with competitive bidding (in the following referred to as auctions ) a are increasingly used for allocating renewable electricity (RES-E) support payments Ecofys Germany GmbH, Berlin, Germany Corresponding author: Malte Gephart, Ecofys Germany GmbH, 10117 Berlin, Albrechtstraße 10c, Germany. Email: m.gephart@ecofys.com

146 Energy & Environment 28(1 2) to RES-E generators (or their intermediaries) and determining the appropriate support level: the Netherlands, France, Italy, the UK, Germany and Spain have recently introduced support schemes with competitive bidding elements for some RES-E technologies; 1,2 other countries are planning to do so in the near future. 2 Overall, a substantial number of European Member States will then use some kind of auction mechanism for allocating RES-E support. In the past decade, most Member States applied administratively defined RES-E support (e.g. feed-in tariffs/premiums), 3 even though some Member States like the UK, Ireland, France, Portugal and Denmark have used auctions for allocating support for (some) renewables in the past (but most of them only for a short period). 4,5,6,7 What is the reason for the European renaissance of RES-E auction schemes? Governments have put more emphasis on controlling support costs and exposing renewables to market principles, which are two elements the European Commission has repeatedly called for, e.g. in its 2013 guidance on support scheme design. 8 Moreover, in April 2014 the European Commission put forward the 2014 2020 state aid guidelines for environmental protection and energy (EEAG) that establish competitive bidding as the key principle for allocating RES-E support and determining RES-E support levels. 9 Member States now have to either use RES-E auctions or justify why they deviate from that principle when registering their support schemes as state aid, resulting in an increasing use of RES-E in Europe. b Despite this recent renaissance of auction schemes, this approach to allocating support for RES-E is not without controversy. The current policy debate on RES-E auctions centres around two main issues. On the one hand, auctions are expected to improve cost-effectiveness and cost control of RES-E support. 10,11 On the other hand, there are concerns whether auctions will be suitable to effectively deliver targeted RES-E deployment levels, 4,12 as experience shows that winning projects may not always be implemented by the project developer. Thus, the question on the effectiveness and cost-effectiveness of RES-E auction mechanism arises. Given the recent introduction of auction schemes and the resulting limits to empirical evidence, the European RES-E policy literature covers auctions for RES-E support less extensively than other options for RES-E policy design, such as feed-in tariff, feed-in premium or quota support mechanisms. Also, until recently, the policy literature that analysed RES-E auction experiences mostly focused on examples outside Europe. 13,14,15,16,17 Recently, a number of European case studies has been published, e.g. by Kopp et al., 18 Held et al. 12 and within the European AURES ( Auctions for Renewable Energy Support ) project. 1,2,5 7 19 27 They analyse main design elements and lessons learnt from past auction rounds in different European countries (see the Literature review section for a more detailed literature review). This article contributes to the policy debate on RES-E auctions by reviewing recent auction literature, with a particular focus on European experience, and exploring which principal design features and context factors increase or reduce the effectiveness and costeffectiveness of RES-E auctions. The article points to trade-offs between different design choices, but also discusses the lack of evidence regarding the impact of some of these design features. Thus, it addresses the question: What determines the effectiveness and cost-effectiveness of RES-E auctions? By identifying some key design options and context factors and discussing them based on empirical examples and theoretical insights, the article provides a unique synthesis of existing insights into auction design, both from theoretical insights and from practical (empirical) experience. It thereby provides support for policy makers for taking evidence-based decisions on upcoming auction designs. Note that this article does

Gephart et al. 147 not compare RES-E auction to other RES-E support instruments and does not recommend one policy instrument over another. Methodologically, the article builds on a literature review and on practical experience in designing auctions. The practical experience includes a study conducted for the German government to prepare and implement the first German pilot auction for supporting ground-mounted photovoltaic systems, 28 as well as a study (and related expert interviews) for the German government to prepare auction schemes for onshore wind, offshore wind and PV rooftop systems. 29 These two studies systematically assessed different auction design options against the criteria effectiveness, cost-effectiveness and actor diversity, building on auction-theory and on interviews with renewables stakeholders, and recommended design specifications for the German PV and wind auctions. The literature review includes publications on international experience with RES-E auctions as well literature that reflects more fundamental auction design options. The article is structured as follows: the following section provides an overview of existing literature on auctions for RES-E. The Basic design elements and key objectives of RES-E auctions section further elaborates on definitions and key design elements of RES-E auctions. It also highlights the different objectives that policy makers may pursue in RES-E auction design, deriving a set of assessment criteria for these objectives. The How to maximise effectiveness in RES-E auctions section analyses key factors and framework conditions that potentially hamper the effectiveness of auctions and discusses design options to improve how effectively auctions can promote the deployment of RES-E technologies. The How to improve cost-effectiveness in RES-E auctions section focuses on how RES-E auctions can support RES-E deployment cost-effectively, first identifying key factors that potentially undermine the cost-effectiveness of auctions, followed by a discussion of design elements that potentially increase their cost-effectiveness. Finally, the last section derives conclusions and recommendations for RES-E auction design in the European context and identifies the need for further research. Literature review As stated above, RES-E policy literature focussing on Europe covers auctions for RES-E support only to a limited extent. However, existing literature provides important insights on fundamental auction design options and on international experiences with RES-E auctions. Key options for and practices in electricity auction design are presented by Maurer and Barroso, 30 who also briefly analyse designs and experiences with RES-E auctions in different parts of the world. They find that auctions foster competition and push prices down, thereby reducing tariffs for end-users and making the whole process more sustainable. 13 However, they also point at two challenges to ensure the effectiveness of auction mechanisms, i.e. the need to attract bidders and ensure competition as well as the need to ensure that auction winners will deliver the awarded projects. To counter the competition challenge, they recommend simple auction processes with a clear set of rules. To counter the delivery challenge, they recommend pre-qualification requirements that discourage speculators or financially insolvent companies from participating in the auction. IRENA and CEM 11 further elaborate design alternatives for RES-E auctions, distinguishing between design alternatives for auction demand, qualification requirements, winner selection and seller s liabilities. Several studies analyse the RES-E auction experience in countries outside of Europe: Kreycik et al. 31 evaluate and compare four procurement strategies for RES-E in the context

148 Energy & Environment 28(1 2) of the USA, namely competitive solicitations (multi-criteria auctions), bilateral contracting, feed-in tariffs, and auctions. They find that auctions perform well with regard to cost-effectiveness but require extensive analysis beforehand by policy makers and regulators; they also conclude that new developers are often not able to qualify or compete in auctions. Rego and Parente 14 and Buckman et al. 32 analyse the auction experience in Brazil and Australia. IRENA 15 and de Lovinfosse et al. 16 compare the design of RES-E auctions in a selection of developing countries. They identify key design elements, relevant context factors, and strengths and weaknesses of the different auction schemes. It is not clear, however, to which extent lessons learnt from these auctions outside of Europe may be applied to the context of the liberalised European electricity markets, in which RES-E is increasingly integrated. Until recently, the empirical analysis on RES-E auction schemes in Europe has been rather limited (see above). Mitchell 33,34 and Butler and Neuhoff 35 analysed the auctions that took place under the Non-Fossil Fuel Obligation (NFFO) in the UK during the 1990s. They observe that RES-E deployment under the NFFO was well below expectations, but their conclusions on the cost-effectiveness of the scheme are more ambiguous. Held et al., 12 Kopp et al. 18 and del Rio and Linares 4 provide an analysis of some more recent RES-E auction schemes in Europe, in particular the experience in the Netherlands, Denmark and France. Del Rio and Linares identify positive and negative effects of auctions, i.e. on the positive side, comparatively low support levels and/or reductions of support levels over time, and, on the negative side, low effectiveness, low technological diversity, modest impacts on the early stages of the innovation process, as well as high transaction and administrative costs. The AURES project provides even more recent case studies on Denmark, 23 Germany, 27 UK, 20 France, 22 Italy, 26 the Netherlands 24 as well a synopsis of lessons learnt. 1 The findings of these empirical studies will be used in the subsequent chapters to provide evidence on the effect of different auction design options. Next to the literature mentioned above that focuses on the RES-E support context specifically, literature on auction theory helps to understand generic choices and effects of auction design. While this article does not provide deeper insights into auction theory, it builds on generic findings by auction theory experts. 36,37 Basic design elements and key objectives of RES-E auctions Basic design elements Strictly speaking, auctions are a mechanism for allocating RES-E support and not a support mechanism by themselves. 13 The main characteristic of RES-E auctions is that they allocate support payments, such as feed-in tariffs, feed-in premiums or capacity payments, c only to a limited number of RES-E projects through a competitive process. Market actors (usually RES-E project developers or intermediaries) bid for a support contract or support right at a price of their choice. The bids that perform best against price, and any further selection criteria defined in the auction notification, are awarded a contract or support right that grants them support payments but also the obligation to deliver the offered capacity or energy volume. Once the budget or volume available in the auction is met, the next expensive bids are not allocated support. The scope of RES-E auctions covers specific technologies (as e.g. in France, Italy and Germany), groups of technologies (as in the UK) or are technologyneutral (as in the Netherlands). Most RES-E auctions in Europe are technology-specific, 1 even though the European Commission pushes for technology-neutral auctions through the EEAG.

Gephart et al. 149 Also, most RES-E auctions in Europe are multi-item auctions, 1 i.e. they allocate support to several bidders/projects, but there are also single-item auctions that select a single winner for specific large projects or sites, e.g. in the case of offshore wind in Denmark, see Kitzing and Wendring. 6 The overall auction volume, which may be distributed over one or several auction rounds, is usually defined by policy makers, e.g. as capacity to be installed (France, Italy, Germany), energy volume (Brazil) or annual support budget (UK, Netherlands). In static sealed bid auctions, bidders submit their bids without any information on the bids of other auction participants. 37 After the auction is closed, the winning bidders either receive the price of their bid ( pay as bid ) or the clearing price ( pay as cleared ). Static auctions are the most common RES-E auction type in Europe. 1 Dynamic auctions are organised in subsequent rounds and allow bidders to react to the information disclosed in the previous round. Dynamic auctions thus reveal more information than static ones. Dynamic auctions are necessarily combined with a clearing price resulting in the same price for all successful bidders. The most common dynamic auction is the descending clock auction; 13 after the auctioneer announces a starting price, auction participants indicate the volume they are willing to offer at the mentioned price. Subsequently, the auctioneer reduces the price in the iterative bidding process until the planned volume is achieved. Brazil uses a hybrid auction scheme that combines a descending clock auction with a static sealed-bid auction: phase 1 operates as a descending clock auction, while phase 2 consists of a static pay-as-bid round for the winners of phase 1. 11,21 At the time this article was written, no European country used dynamic auctions for RES-E, even though the sequential rounds of the Dutch SDE+ (Stimulering Duurzame Energieproductie) auction scheme include elements of an ascending clock auction, see Held et al. and Noothout and Winkel for a description of the Dutch SDE+ scheme. 12,24 The choice of the auction type and clearing rule is only a small part of the overall auction design. Important elements of the auction design include the choice of the auction product, material and financial qualification requirements (also called eligibility requirements), requirements for financial guarantees, delivery periods and/or penalties, as well as the auction schedule and frequency. 1,11,29 Ceiling prices that define the maximum possible support level can limit the risk of high support costs for governments/consumers in case of insufficient competition. 1 Auctions may allow for the return or transfer of support contracts or support rights, which may also influence the auction result. 29 The article elaborates on the role of these provisions in the subsequent chapters. Another fundamental question is what type of remuneration is allocated through the auction (e.g. feed-in tariff, feed-in premium, capacity payment, etc.). However, this latter question is rather an issue of general support scheme design that has been discussed extensively (see e.g. Winkler et al. 38 ) and will not be further discussed here. Most European auction schemes provide access to (sliding) feed-in premium support, 1 as the EEAG request that production support to renewables is paid as a premium on top of the electricity market prices, thus excluding feed-in tariff support, except for small-scale installations. Key objectives of RES-E auctions It is important to understand that the design of an auction scheme also depends on the policy objective behind the auction scheme, which will be reflected in the auction design. 28

150 Energy & Environment 28(1 2) Important objectives of introducing RES-E auction schemes in the European countries have been cost-effectiveness and control of overall support costs. 1 According to the IPCC, economic effectiveness relates to the the impact of policies on the overall economy, which in turn includes the sub-criteria economic efficiency and cost-effectiveness. Economic efficiency relates to the allocation of resources, which is efficient if it is not possible to reallocate resources so as to make at least one person better off without making someone else worse off (known as the Pareto criterion ). 39 42 Cost-effectiveness is described by the IPCC as achieving a specific objective at the lowest cost. 39 Adopted to the field of auctions for renewables support, cost-effectiveness can be understood as the achievement of a given RES-E target at the lowest possible cost to society. 43,44 In addition to support costs, cost-effectiveness includes cost elements such as transaction costs for implementing a support scheme or the costs of physical integration of RES-E (grid and backup costs). In practice, transaction and system integration costs are difficult to quantify. For this article, we limit the definition of cost-effectiveness to the support costs for reaching a specific RES-E target, which is also a common concept among policy makers. A basic objective of any policy is effectiveness; in the context of RES-E, auctions is their effectiveness in deploying RES-E and reaching RES-E targets. Other common objectives of environmental policies include distributional fairness (equity), and institutional and political feasibility. 39,40,45 48 RES-E policies may also pursue further specific objectives, e.g. exposing RES-E generators to market principles (and complying with the EEAG), avoiding market concentration, ensuring market access for smaller market actors, etc. IRENA and CEM 11 highlight the maximisation of socio-economic benefits as one important policy goal of policy makers and identify different mechanisms that require auction winners to play an active role in regional development, such as local content requirements and requirements regarding empowerment and employment. In European RES-E auction schemes, such provisions tend to be less prominent than in some developing countries, as they may conflict with European and WTO trade laws. Nevertheless, different examples show that socio-economic considerations play an important role in European countries. For example, the German wind auction scheme that will enter into force in 2017 grants community projects less stringent prequalification requirements. The French auction for medium-sized PV evaluates bids not only based on price but also based on CO 2 footprint, which favours French over German and Chinese PV modules, and may thus be interpreted as implicit local content rule. 22 Some of the above-mentioned policy objectives may turn out to be in conflict with each other, therefore auction design inevitably involves trade-offs. In the following sections, the article analyses which auction design elements can help to achieve the policy objectives of effectiveness and cost-effectiveness of RES-E support, also exploring the trade-offs between the auction design options for reaching these objectives. How to maximise effectiveness in RES-E auctions The risk of low realisation rates Volume control of incentivised RES-E deployment is a key goal of auction schemes. 49 By setting the annual auction volume and frequency, the auctioning authority exerts per se high control over the maximum RES-E deployment. However, the effectiveness of auctions, thus the actual deployment level, has varied significantly across international auction schemes. 1,12 Yet, the authors of this paper assume that, for policy makers effective auctions are important

Gephart et al. 151 for credibility of support, for the public acceptance of auctions and for attaining RES-E targets. Data on implementation rates of RES-E auctions is somewhat scarce because in many recent auctions, the implementation period of the awarded project has not yet expired. Available case-by-case insights suggest that implementation rates have usually been below 100% and delays are frequent. 1 Moreover, implementation rates have been mixed between and within auction schemes, resulting in lower predictability regarding the actual deployment of RES-E. While all auctioned Danish offshore wind projects have now been realised, 6 implementation rates of the Non-Fossil Fuel Obligation (NFFO) in the UK between 1990 and 1998 were at 38% of selected projects and only 26% of the contracted capacity. 34 More than half of auctioned projects in France (solar PV 100 250 kw, 2011 2014) and Italy (onshore wind, 2013 2015) had not been finalised by their deadline. In the Netherlands SDE+ scheme (2011 2015), while only around 30% for geothermal heat projects auctioned in 2011 have been realised, about 70% of onshore wind auctioned projects have been built. 1 In Brazil, only 30% of the onshore wind capacity contracted in the first three rounds (2009 2010) had started operating on time, and 40% suffered delays greater than one year by the end of 2013. 17 A limitation in the assessment of the realisation rate is that some projects are realised, but only after the formal realisation deadlines. The difference in implementation rates can be explained by specific fail factors and by analysing whether these fail factors have been eliminated in the auction or not. In other words, find out whether auction participation is limited to bidders that have the capacity to deliver the quantity of energy promised in the contract in a timely manner. 11 Typical fail factors of auction bids include general risks related to project development, actor-specific risks and risks related to (unwanted) strategic behaviour (especially underbidding). Risks related to general project development which do not uniquely emerge in auctions but in RES-E deployment under all support schemes may include that a project did not get planning consent or that a project is not connected to the grid (or the latter being delayed). 29 A supply chain that is lacking can also be cause for project delays or failures. 1 As the bidder does not know whether his project will be awarded support in the auction, he will attempt to minimise the investments into the project development, as theses will be sunk costs. d Consequently, these project-related risks may still exist after the project has been awarded, potentially resulting in project failure. Other risks relate to the technical and financial capability and capacity of the project developer: 11,29 a project developer may not have sufficient experience, resulting in delays or cost increases which make the project economically unfeasible. A project developer may not have secured sufficient funds for his project and as a consequence may face financing problems at the realisation stage. Other reasons for project failure are related to speculative strategic behaviour by auction participants and to the so-called winner s curse. Underbidding, i.e. handing-in an unprofitable bid price, can occur knowingly or due to lack of experience. Bidders may seek to strategically influence the auction result through deliberate underbidding, although they might not have the actual intention to implement awarded projects or they might crossfinance projects (such as in the case of cross-financed state owned enterprises in the Chinese Wind Power Concession Programme from 2003 to 2007). 50,51 Underbidding occurs because auction participants might want to crowd out competitors, thereby increasing chances of their projects being awarded and/or to be able to bid with higher prices. Underbidding can

152 Energy & Environment 28(1 2) also occur unintentionally when bidders are not sufficiently familiar with the auction process or underestimate their own project costs ( winner s curse ). Design options to increase realisation rates There are a number of design elements of auction schemes that address these risks. The most important design elements are material and financial qualification requirements, predefining the period until project completion, and penalties. Material qualification requirements address project risks and can increase project implementation rates by limiting bids only to projects that already demonstrate an advanced project development stage. 1,29 These include, for instance, demonstrating evidence of planning consent or demonstrating evidence of a grid connection agreement. Most European auction schemes analysed by the AURES project 1 used material qualification requirements, typically a building or planning permit and/or a grid connection agreement. The only auction that did not require any material qualification requirement was the Spanish wind auction in early 2016, which may explain why the auction winners were outsiders not active in the wind sector so far. 19 The Netherlands have increased the qualification requirements of the SDE+ scheme: Since 2014 it demands next to other requirements the submission of a feasibility study for projects larger than 500 kwp or 50 Nm 3 /h. 24 In the UK, the NFFO auction did not define any qualification requirements, which can be seen as one reason for its low implementation rate. In contrast, the 2014 auction under the UK Contract for Difference scheme defined relatively stringent prequalification requirements. 20 Material qualification requirements help to reduce project risks which can lead to project failure after the auction. However, effectively, qualification requirements shift project risks from the auctioning authority to the project developer the more demanding the qualification criteria, the more risks are handled by the project developer before even participating in the auction. Material qualification requirements necessitate investments in the project progress which represent sunk costs. Such costs increase the bidding risks and can discourage potential bidders from participating in the auction. This can limit the competition in the auction round. Material qualification requirements can also increase the administrative costs of the auctioning authority for assessing the fulfilment of the qualification requirements. Other possible qualification requirements include proofs of the bidder s financial or technical capability (IRENA and CEM 2015), thereby addressing actor-specific risks. Financial qualification requirements can include bid bonds that need to be submitted together with the bid or after the awarding and which are reimbursed after project completion. 11,28 Technical capability can be ensured by proving, for instance, a certain amount of years of experience in similar projects. Bidders in Denmark (offshore wind) and Portugal (onshore wind) were required to provide proof of their experience building and/or operating wind farms. Bid bonds are a common requirement in many auction schemes, both inside and outside Europe; for instance, Denmark, Germany, Italy, Portugal, Brazil and South Africa have introduced bid bonds to ensure bidders serious intent to realise projects. In Germany s solar PV pilot, 52 auction participants had to provide a first bid bond to be considered, followed by a second bid bond in case of being awarded a contract. The value of the bid bond depended on whether bidders provided a preliminary zoning plan (4E/kW and 50E/kW, respectively, for the first and second bid bond) or a definitive zoning plan (2E/kW and 25E/kW, respectively, for the first and second bid bond). 1 In Italy, bidders have to hand in 5% of the administratively estimated investment cost per technology upon application. 26

Gephart et al. 153 Predefining the period until project completion is another important design element that in combination with qualification requirements and penalties affects the effectiveness of an auction. 1 If the implementation period is short, this increases the risk of delays. On the other hand, if combined with sufficient penalties, short implementation periods also incentivise quick project realisation and implicitly force bidders to participate with projects at advanced development stages (such advanced development may also be proven explicitly by material qualification requirements, see above). One can thus distinguish between late auctions, which are characterised by short implementation periods and advanced project progress of participating bids, and early auctions, which require little pre-development and grant longer implementation periods. Most European auction schemes analysed within AURES 2 are late auctions. 1 France (solar PV 100 250 kw) and Germany (wind onshore) conduct late auctions and require projects to be built within 18 months and 24 months, respectively, after being awarded a contract. That is to say, countries also set realisation periods depending on the technology: developing an onshore wind project takes considerably longer than a PV project. There are different types of financial and non-financial penalties to address the risk of unwanted strategic behaviour, incentivise a quick project execution and to avoid non-implementation of projects. 1,11,28 If projects are delayed, support levels can be lowered, thereby reducing the RES-E generator s remuneration (and thus the return on investment). Alternatively (or in combination), the duration of support can be reduced, e.g. by subtracting twice the time span of the delay from the overall support period like in the French PV auction. 22 If project delays exceed a certain time period, the ultimate penalty can be the elimination of the support right (or the termination of the contract, respectively) and the exclusion from future auction rounds (e.g. applied in the Netherlands and the UK). Another commonly applied penalty option is a financial penalty, i.e. a fine to be paid if a project fails. Financial penalties are usually guaranteed through bid bonds (i.e. financial deposits or guarantees) in order to avoid bidders circumventing their liability through insolvency of the project company. Such penalties apply, e.g. in Germany, Denmark and Italy. 1 However, small bidders in particular might not have sufficient credit-worthiness to hand in the bid bond and therefore might not be able to participate in the auction. 29 Based on talks with financial institutions in Germany, they might still be able to obtain the required guarantees by demonstrating an advanced project development stage, but this will depend on the maturity of the technology market, the experience of the bidder, and the risk aversion of the bank. In Germany, this was a reason to stretch the bid bond over several milestones in the project development for community projects. 53 It is important to note that unlike material qualification requirements which ask for specific proof on the quality of the bid, penalties grant the bidder more freedom in regards to which measures to take, e.g. selection of adequate sites and securing permits, to ensure project implementation and avoid penalties, or securing funds for potential penalty payments. 28 Moreover, compared to material qualification requirements, penalties lead to lower sunk costs for participating in the auction, but they also increase the bidding risks and thereby increase risk premiums. Other options to address project development risks and project failure are to make the support right (or contract) transferable or to re-auction the volume of the failing or delayed projects. Such provisions may ensure that non-implemented projects can be replaced by other projects. 28 This way, the targeted RES-E volume may still be met, despite some

154 Energy & Environment 28(1 2) delay. Conceptually, such provisions may be of special relevance in early auctions with no or few material qualification requirements. In practice, such provisions are hardly used or appear to be of limited importance in the analysed countries that mostly use late auctions. One option that was applied in the PV pilot auction in Germany is to allow bidders to return the support right/contract earlier in exchange for a smaller penalty. In this case, the auctioning authority can re-auction the support right/contract and replace non-realised projects in a shorter time frame than if the entire time period for project implementation had been exceeded. In practice, this option has never been used in Germany (as of July 2016). Instead of a re-auctioning of support rights/contracts, an automatic replacement system where the support right/contract is automatically transferred to the next highest bid can also be considered. An automatic replacement could, however, potentially prolong the auctioning process and force bidders to sustain their bid for longer periods of time while their permits expire (this has been problematic in Denmark, where the 2009 Anholt tender included an automatic replacement, but no extension of the implementation deadline). 6,12 Instead of handing support rights back to the auctioning authority (or terminating the contract), a transfer of the support right/contract to other projects of the same bidder or sale of support rights to other market actors could be allowed. 28 In this case, the bidders that were originally awarded the support right/contract could transfer the right/contract including the associated obligation to implement the project. Depending on the circumstances, this could result in a secondary market for support rights/contracts and further increase flexibility for bidders. RES-E producers would be offered the opportunity to purchase support rights/contracts without participating in the auction. However, the transferability of support rights/contracts may also lead to unintended effects and speculation. For instance, project developers may refrain from participating in an auction, aiming instead for support rights/ contracts that they can purchase from an actual auction participant. This actual auction participant might sell its support rights/contracts at preferable conditions to the off-taker, if he faces project delay or failure, in order to avoid costlier penalty payments. The transferability of support rights/contracts is an interesting option to increase project realisation rates, but is hardly applied in practice. The German pilot auction for PV allows the transfer of the support right to other projects of the same bidder against a small penalty, but no data are available on the extent to which this provision is used and on its effectiveness in ensuring project implementation. In discussions with European wind project developers, many of them voiced concerns that transferability would lead to speculation and participation of unexperienced bidders, thus reducing the chances of serious bidders. 29 Different types of qualification requirements and penalties are in principle all suitable measures to improve the effectiveness of auctions. The actual design choice appears to be rather context-specific, but the evidence available from European and non-european countries points towards a preference for late auctions with stringent material qualification requirements, despite a number of drawbacks. 1 There is far less experience with the transferability of support rights. High effectiveness is likely to come with a price as high qualification requirements will limit the degree of competition and bidders will price in penalty risks (even though there is no empirical data to quantify this effect). Thus, a balanced approach between ensuring high realisation rates and potential for low clearing prices needs to be struck, well able of reflecting the technology-specific project development cycles, costs and risks.

Gephart et al. 155 How to improve cost-effectiveness in RES-E auctions The aim of improving cost-effectiveness and mixed empirical results Improving the cost-effectiveness of RES-E support is a major objective of auctions. 4,54 This can happen in two ways. First, as the European Commission 4 argues, auctions can lower revenues for RES-E producers as much as possible. Auctions circumvent opportunities for lobbying, which might occur when support levels are defined administratively. 12 This is seen as one of the main reasons for high support-levels and thus for excessive revenues for RES-E production. Moreover, auctions can induce competition among its participants and can incentivise project developers to reveal their actual costs. This potentially reduces the cost per unit (e.g. kwh) in comparison to an administratively defined support level. Second, auctions potentially select the cheapest options to produce RES-E, thereby lowering the overall support costs. 4,11 This can, for example, happen in technology-neutral auctions by selecting the cheapest technology (albeit with drawbacks regarding the complexity of auction design and dynamic efficiency). e Moreover, auctions that do not predefine locations lead to an improved selection of cheaper sites in terms of resource availability. As a result, installations will be built where the respective resources such as sun and wind are best available. However, locational signals or specific eligibility criteria might be required to avoid adverse effects on the overall cost-effectiveness on the system level (e.g. when site selection according to resources ignores system integration costs). Auctions are theoretically capable of reducing support costs by reducing support level, which is used in this article as one indication of increased cost-effectiveness. But how have auctions performed empirically in terms of prices? These experiences have been analysed in a number of publications and have shown a broad range of results. 1,12,15,34 Some auctions have produced significant reductions of support levels in comparison with prior existing feed-in tariff systems or compared to auctions in other countries. Most prominently, auctions in Brazil have led to a 60% reduction in the support level for wind onshore, from an average of E135.7/MWh under the previous feed-in tariff scheme, to an average of E54/ MWh in 2009 2013. 17 But in Germany, where average support levels granted for groundmounted PV decreased from E91.7/MWh to E72.3/MWh within one year (figures including electricity market price), the auction scheme is also seen as effective in bringing down PV cost. 2 However, other auctions have produced higher-than-expected support levels, such as medium-scale PV auctions in France (2012 2014), resulting in an average price of E193.4/ MWh or E20 30 above that paid under the previous feed-in tariff. 22 The fact that the amount of eligible bids in the 2013 2014 rounds was less than 1 MW above the auctioned volume, indicates that there was barely any competition, which may have compromised the cost-effectiveness of the auction. However, in 2015 the demand strongly increased to 2 GW of bids for 200 MW of auctioned capacity, which indicates competition levels have improved. It is important to note that these tariffs cannot directly be compared to each other, as some of these auctions are technology-specific and others are technology-neutral (and even include RES-heat, as in the Netherlands). Moreover, they cannot be directly compared to administratively defined feed-in tariffs. In a feed-in tariff system, only commissioned projects receive the support. In this support scheme, projects that have been planned but have not been implemented do not enter into the analysis of the support scheme s cost-effectiveness. In contrast, in an auction the focus usually lies on immediate auction results, regardless of whether a project will be realised or not. Thus, comparing administratively defined support levels with those in auctions is somewhat biased. 35

156 Energy & Environment 28(1 2) Design options to improve the cost-effectiveness of auctions What are the principle design elements that influence the cost-effectiveness of auctions? The paper focuses on a selection of design elements that have been identified as the most relevant: 1,11,12,29, first, the level of competition induced by the auction; second, the minimisation of unwanted over- or underbidding; and third, the reasonable limitation of risks for project developers/bidders (or the adequate distribution of project risks between project developer and auctioning authority to limit the related risk premiums). A good auction design addresses these requirements and at the same time takes into account numerous context factors, such as the overall market situation, the technology segment-specific market potential, the project development risks, and the structure and characteristics of potential bidders. In turn, the auction design partially defines the relevant market segment and the accessibility of the auction (and thus the level of competition). First, the level of competition in an auction greatly determines its outcome. The higher the number of participants, the more pressure on them is increased to lower prices. The level of competition depends on the broader context, i.e. the RES-E market environment in a country. 28 In addition, the regional and even global economic situation matter for the outcome of auctions. If in one regional market (e.g. Europe) RES-E growth is decreasing, auctions in other more promising markets might attract more international investors. This effect seems to be one of the reasons for extremely high levels of competition (and resulting low prices) in auctions in Brazil. 21 Negative effects of international competition could be observed in one of the Danish offshore auctions. 6 An auction design that avoids excessive risks (or potentially sunk costs) for project developers results in an accessible auction, invites more actors to participate and thus leads to increased competition among participants. Keeping qualification requirements as well as penalties at a reasonable level improves the level of competition in an auction (see the How to maximise effectiveness in RES-E auctions section and below). Moreover, the auctioned volume is a crucial factor that influences the cost-effectiveness of an auction. 11,13 In a functional competitive bidding process, supply excess of RES-E projects in relation to the auctioned volume is a prerequisite, meaning that more capacity needs to be available in projects than support can be allocated (scarcity requirement). The degree of excess supply depends on the auctioned volume in a given market context. The frequency of auctions is also important. 1,28 The entire annually available volume might be auctioned in a single auction which would lead to relatively long time periods between the auctions and related insecurity for smaller project developers. On the other hand, the annually available support might be divided into several auction rounds, leading to a more continuous access to support. Whether support rights are made transferable is another design option that does not only influence the effectiveness of auctions, but also their cost-effectiveness (which is why the design option is briefly discussed again here). In terms of cost-effectiveness, such transferability leads to greater flexibility among bidders. Allowing the transfer of support rights might create a secondary market in which intermediaries, e.g. financial institutions, could also participate. This might reduce costs and thus required support levels because in a secondary market, bidders can hedge against general project development risks (and related penalty risks). 28 On the other hand, it can also lead to higher support levels in the primary auction compared to the secondary market. 28 However, there is limited experience regarding

Gephart et al. 157 the effects of secondary markets on the cost-effectiveness of auctions, as such provisions are hardly used in the countries analysed. Apart from these more general issues that determine the degree of competition in an auction, the mitigation of the winner s curse and unwanted strategic behaviour resulting in artificially low or high prices (i.e. not close to project costs) is a crucial element for costeffective auctions. Important elements that influence the behaviour of auction participants are the auction procedure together with the pricing rule and potential ceiling prices. The following paragraphs will briefly discuss a sealed-bid auction versus a descending-clock auction and the pay-as-bid pricing rule versus uniform (pay-as-cleared) pricing. A sealed-bid auction procedure is implemented in Peru and South Africa 16 and in the pilot PV auction in Germany. 28 The advantage of the sealed-bid auction design is that bidders are less likely to act collusively and thus increase prices by illegitimately coordinating their bids compared to a dynamic auction design. Instead of a sealed-bid auction, a dynamic descending clock auction can be applied (e.g. as in Brazil, see above). 12,13,16 A descending clock auction with uniform pricing can produce valuable information for the auctioneer and for market participants in a context of high uncertainty regarding the cost of RES-E. In the course of the auction, participants learn how their competitors behave and can judge the level of competition. The advantage compared to a sealed-bid auction is that bidders might lower their bids to the minimum price feasible if competition is intense. The disadvantage is that bidders are more likely to act collusively and thus increase prices by coordinating their bids. 37,55 Regarding the pricing rule, the pay-as-bid pricing approach is a very simple and easy-tounderstand process, which is specifically suitable for less experienced auction participants. 12,28 The downside of this pricing approach is that participants will potentially guess the level of competition and will, if competition is expected to be low, bid above their actual costs, thereby potentially lowering the cost-effectiveness of the auction. Unlike the pay-as-bid pricing rule, a uniform pricing approach can lead to low tariffs, because for the bidder bidding his real cost tariff does not lead to a disadvantage. 12,28 As the single participant will usually not set the price he receives at the auction, there is no immediate need to strategically bid higher or lower than the actual production costs would indicate. In practice, there have been some cases in which bidders submitted unrealistically low bids, e.g. for solar in the UK, 20 possibly because bidders were unexperienced. Furthermore, if in an auction with uniform pricing bidders can submit multiple bids, they might be able to influence the overall outcome of the auction by driving up the clearing price through additional bids. 28 The UK addresses this problem by setting out requirements on how these multiple bids need to be designed (e.g. different volume, different delivery years and different prices). Alternatively, a pay-as-bid pricing rule might be chosen in the case of multiple bids, as the auction participant cannot influence the entire auction result (i.e. the price for all participants) through his bids. A ceiling price can eliminate the risk of excessive support levels in a (pay-as-bid or pay-ascleared) sealed bid auction: it excludes all bids which are above a certain threshold. However, if ceiling prices are not disclosed, they can result in many bids being excluded from the auction, because they exceeded the ceiling price. As a result, the auctioneer might fall short of his envisaged auctioned volume. If the ceiling price is disclosed, this problem is circumvented as participants know the support level that will exclude them from being selected. However, if competition is low, auction participants are likely to bid close to the ceiling price. 28 In case of sufficient competition, a ceiling price has no negative effect.

158 Energy & Environment 28(1 2) Table. 1 Summary of design elements that influence effectiveness and cost-effectiveness. Auction design element Impact on effectiveness Impact on cost-effectiveness (Material) prequalification (++) Pre-selection of strong requirements projects/bidders increases effectiveness Bid bonds and penalties (++) Penalties for non-realisation increase effectiveness Delivery periods (+) Adequate delivery periods need to be defined to ensure effectiveness; Short delivery periods imply pre-development of projects Transferability of support (no evidence) (+) Unsuccessful right or contract project can be replaced by new projects; (-) Implies less developed projects and delays in implementation Auction volume (+/ ) Auction volume should not exceed demand for support; Predictable auction volumes improve project pipeline and effectiveness Auction frequency (+/ ) Frequency should be aligned with project development cycle Auction type and price rule (+/ ) Most effective auction type and price depends on market context Ceiling price (+/ ) Increases stability of auction scheme by avoiding excessive support costs; If ceiling price is set too low, auction volume might not be met ( ) Pre-selection of projects/bidders limits competition ( ) Penalties increase risk premiums for bidders and limit competition (+/ ) Short-delivery periods imply delivery risks and sunk costs; (no evidence) (+) Reduces risk premiums ( ) May lead to unwanted strategic behaviour and higher prices (+/ ) Auction volume needs to be scarce (i.e. below volume of projects) to create competition (+/ ) High frequency avoids stop and go and reduces bidder risks but increases risk of collusion (unwanted strategic behaviour) (+/ ) Most cost-effective auction type and price depends on market context (+) Eliminates the risk of excessive support levels In case of low competition bidders may bid close to the ceiling price, but no negative effect in case of competition Other important elements that influence the cost-effectiveness of an auction are different risks perceived by project developers and investors. In this respect, a trade-off emerges when implementing auctions for reasons of cost-effectiveness. The cost of capital constitutes a significant element of RES-E development costs. Cost of capital is related to the perceived risk of an investment, which in turn is heavily influenced by policy and regulatory risks. 56 This article assumes two type of risks to be particularly relevant in this context: 12,29,57,58 first, the risk of not being granted access to support for a project (allocation risk) and, second, the risk of non-implementation of a project (delivery risk), which might result in penalties for the bidder. Regarding the first risk, projects might need to be developed to a certain degree at the time of participating in an auction. This is necessary to meet potential