Collecting financing data across the EU
The cost of capital data was collected through structured interviews with stakeholders in EU Member States. The research team conducted the majority of the interviews in the local language of the respective country, or in English as default. The team consisted of Master and PhD students across the European Union, which cover all the official EU languages, and consisted of 18 members. A five-step methodology was followed in collecting the data, as explained below.
Defining the scope. First, a distinction was made between focus and non-focus countries. When conducting the interviews, the research team used two different survey templates. On the one hand, for non-focus countries, the template was limited to obtain quantitative data on specific financial variables for renewable energy projects. On the other hand, the template for focus countries was designed to allow for more in-depth research. Hence, it included not only the quantitative questions on financial variables but also other qualitative questions to trigger discussions about the evolution of the costs of capital and the impact of auctions. Both templates are available in the extended data (Brückmann et al., 2021).
Five EU Member States were selected as focus countries: Denmark, Germany, Greece, Portugal, and Spain. When selecting the focus countries, an energy policy exclusion criterion was taken into account: only countries that had conducted at least one renewable energy auction were considered. The reason behind this criterion is that one of the objectives of the AURES II project is to understand the interrelations and inter-dependencies between auctions and financial variables.
At the moment of selecting the focus countries, the following Member States had already conducted auctions: Croatia, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Lithuania, Luxembourg, Malta, Poland, Portugal, Slovenia, Spain, the Netherlands, and the United Kingdom. From these countries, the research team selected the focus countries considering a geographical balance.
Identifying the interviewees. Second, based on desktop research and experience from previous similar projects, different relevant stakeholders were identified and categorized: banks, project developers, financial experts, utilities, investment funds and others. Although the list of stakeholders involved in the energy transition is much broader, the scope of WP5 was limited to collect project-based quantitative and qualitative cost of capital data, which ultimately defined the categories of actors. The main criteria in including a stakeholder in the research was direct experience with actual projects and their financing - nevertheless, we did not apply any formal criteria, for instance a minimum number of years of work experience. The main reason behind this is practical - interviewee acquisition is challenging, especially if the topic is sensitive, like in our case. Auctions are based on price competition, where costs of capital are key information in final bid formation. Therefore, revealing this to the competition could cause direct financial damage, for instance in the form of sunk costs because of losing a bid. We agreed on the above defined categories based on project-internal review rounds during the development of the survey. Therefore, these categories were verified by other consortium members.
We mapped the stakeholder categories along the project development timeline. During project development, professional project developers conduct the initial economic and technical studies and lead the project through the permitting process. In some cases, larger utilities or energy companies conduct this activity in-house. To decrease technical risks during construction and operation, insurance companies provide projects with insurance packages. These sometimes also hedge projects against price risks, especially if the projects’ remuneration scheme and support contract expose the project to volatile electricity markets. After the project wins support in an auction, bankers finance the projects, typically in project financing deals. When the project is constructed and operational, larger institutional investors and/or private equity funds buy shares in the project - for instance, a pension fund seeking to "green" its investment portfolio. These investors may also finance projects during their development stage. When a stakeholder did not fall within any of these categories, we mapped him or her under the category "other". For example, we interviewed many consultants that provide financial advice but do not fall within any of the above listed categories.
To select the interviewees, the research team conducted extensive desktop research to reach possible interview partners including different kind of stakeholders involved in the financing of wind and photovoltaics (PV) projects, e.g., investors, developers, banks as well as renewable energy associations. This was made by identifying two groups of countries. Countries with wind and PV auctions between 2018 and 2019 (all EU and UK, except for Austria, Belgium, Bulgaria Estonia, Ireland, Croatia, Latvia, Romania, Sweden, Slovakia, Czech Republic and Cyprus) and countries without auctions at that time. In countries where PV and wind auctions have been conducted, possible interview candidates were identified by screening the publicly available lists of auction participants, as well as their renewable energy associations. In these countries, the renewable energy association helped to identify further stakeholders involved in finance of PV and wind projects. In countries where auctions had not taken place yet, the renewable energy associations were contacted to obtain data, as well as to get support to identify potential interviewees involved in finance of wind and PV. Then, the contact information was extracted from the potential interviewees’ websites, including e-mail and phone (in case this information was already available on the organisations’ websites).
Previously, some consortium members were involved in two EU-funded projects (DiaCore1 and Re-Frame2), for which similar data was collected. The professional networks built in these two projects were also used to arrange interviews for WP5 of the AURES II project.
Designing the survey template. Third, the consortium designed and elaborated the questionnaires to be used in the interviews, both for focus and non-focus countries, which can be found in the extended data (Brückmann et al., 2021). To ensure that the survey asks relevant questions and is structured in a comprehensive and logical manner, we conducted five initial exploratory interviews and adjusted the survey based on the received feedback (the original version of the questionnaire used for the exploratory interviews can be found in the AURES II exploratory interviews template Brückmann et al. (2021)).
The questionnaires start first with information on work background and experience of the interviewee, followed by questions on quantitative values for financing variables of RE projects: WACC, cost of debt (CoD), cost of equity (CoE), debt service coverage ratio (DSCR), loan tenor, technology, size of the project, type of financing and time of completed construction for solar PV, wind onshore and wind offshore. The questionnaires asked for values of specific projects. However, if the interviewees did not want to reveal data that could lead to the identification of a concrete project, they were asked to provide general country estimates instead (i.e., an average WACC value for instance, which can be an average estimate of RE projects in the country and in the time considered).
In the five focus countries, we complemented the quantitative section with a semi-structured discussion on the impacts of auctions on the financing variables, the reason behind the changes in their values, and how specific auction design elements affect project financing (see the focus country questionnaire in the extended data (Brückmann et al., 2021)). First, the interviewees were asked if the cost of debt, cost of equity, DSCR, loan tenor, and debt to equity ratio changed after the introduction of auctions and why these changes occurred. Second, they were asked to provide the main three reasons of the changes and to also rank the reason(s) according to their importance between 0 and 4, where 0 means not important at all (i.e., not applicable), 1 slightly important, 2 important, 3 fairly important and 4 very important. To stimulate a discussion, they were shown an illustration (see focus country questionnaire in Brückmann et al. (2021)) that depicts different auction design elements (such as auction volumes, bid bonds, and penalties), and connected them to different stages in the lifetime of a renewable energy power plant. For instance, auction volumes and frequency may affect planning risk during the pre-development and development phases. Material and financial pre-qualification requirements may affect the bidders’ allocation risk (the risk of obtaining support or winning an auction) - stricter requirements increase potential sunk costs. Penalties and realization deadlines could affect the risk of being penalized, while remuneration scheme design affect directly the project revenues and may impact credit risk - namely the debt providers risk of recovering the loan.
Conducting the interviews. Fourth, between November 2019 and April 2020, the consortium conducted interviews in EU Member States (and the United Kingdom), consisting of one interview session per each interviewee. All interviews were conducted via online conference tools or through a telephone call. The average duration of the focus countries’ interviews was around 45 minutes, whereas the average duration for the non-focus countries was around 25 minutes. The templates used for both focus and non-focus countries can be found in the extended data (Brückmann et al., 2021).
The research team conducted the interviews individually and in groups of two researchers, where this was possible. Due to the large scope of the survey involving all EU Member States, we did not manage to ensure the consistent participation of at least two interviewers. We acknowledge that this is a limitation, as those interviewers that conducted interviews individually might have misinterpreted some answers. In the majority of cases, interviews were not recorded due to confidentiality reasons and the interviewer took personal notes during the course of the interview. However, in few cases, the interviews were recorded (with the prior consent of the interviewees) and the recordings were used to re-listen to the discussion and to note down the main points. In this way, the interviewer had more capacity to engage in the discussion without taking notes. The recording files were automatically deleted afterwards. The overall notes were then used to explain the quantitative survey findings - published in a project report (Roth et al., 2021).
Arranging interviews and accessing financial variables was challenging because these data are key for private actors and may be considered as a trade secret. Hence, in order to mitigate this challenge, the interviews were conducted under the Chatham House Rule, meaning that the interviews were anonymous and the interviewees’ answers cannot be used to identify them. The details of the participants were never published or disclosed to third parties (inside the AURES II consortium as well as external parties).
The approach used to arrange and conduct the interviews was based on different steps. First, when the researchers contacted the potential interviewees for the first time, they were asked for their consent to participate in an interview concerning the aforementioned financial variables, highlighting that the interview would be anonymous. In the e-mails with the interview invitations, a .pdf document was attached explicitly mentioning that the interview would be anonymized, and that “your answers will never be linked to your identity or your organization” and also that “you can withdraw during or anytime after the completion of the interview before the publication of the study”. Besides, a link to the official website of the Chatham House Rule was included in this document.
During the interview we explicitly noted that the consortium will use all information/data for the purposes of the AURES II project but keeping their identity confidential. We decided not to collect the written consent forms since in most cases interviewees were not willing to sign and send them back. Nevertheless, participants were asked orally before the start of the interview whether they agree to those.
In addition, during the interview, it was again stated that interviewees can withdraw during or any time after the completion of the interview and before the publication of the results.
Ethical approval was not required for this study. Original interviews containing all the data, i.e., the completed questionnaires, have been internally stored on eclareon’s server to prevent access from third parties. Only anonymized versions of interviews were uploaded to the AURES II consortium platform (which is only accessible to AURES II consortium members). Due to confidentiality concerns, we do not include any transcripts and notes in the extended data3.
Data collection and processing. Fifth, all collected data was centrally extrapolated and compiled into an Excel file. To process the data, we used Excel 2016 in these steps.
First, we developed an input template, where each column represents a single question or data value (for instance WACC values), while each row represents one response. To anonymize the data, the responses were given codes - for instance in Austria we collected 11 responses and named them from AT01 to AT11.
Second, instead of providing a specific value for the financing variables, some interviewees gave us a range of estimates, as they were responding to the survey from memory and not by reading a specific value from a financing term sheet. Consequently, we created subcategories that differentiate between the best and worst input per each financing variable and per each survey input. We assigned these sub-values the codes such as "AT01 - best fin" and "AT01 - worst fin". To maintain consistency and structure across the dataset, we did this for all the inputs - regardless if they are a single estimate or a range estimate.
In a third step, we cleaned the input data so that it does not contain space values and that the inputs are inserted using the same number format and/or words (for instance "onshore wind" instead of "wind onshore" or just "onshore").
Fourth, we summarized each financing variable into minimum, maximum and average data input per country and technology. We did not calculate median and upper/lower quartile values and outliers since the dataset contains a limited number of inputs per country and technology.
Finally, we excluded from the available dataset the country-technology cases for which we have collected less than 3 survey inputs. We do this to maintain data confidentiality, as the low number of data inputs could enable the users of the Data Note to identify a survey respondent. This led to the exclusion of the following country-technology cases: wind onshore in Bulgaria, Cyprus, Hungary, Ireland, Slovakia, Slovenia, and the United Kingdom; solar PV in Bulgaria and Cyprus; and wind offshore in Ireland, Latvia, and the Netherlands.
Although we have collected 240 survey inputs, we show in the available dataset (Brückmann et al., 2021) the information for only 206 estimates, because of this exclusion criterion4. While this decreases the number of countries the Data Note covers, it increases the robustness of the data and we can ensure the confidentiality of the interviewees.
