Real options applications to assess the impact of public funding on r and d projects from a double perspective: the private company and the funding body

Real Options applications to assess the impact of public funding on R&D projects from a double perspective: the private company and the funding body By: Gonzalo Martín-Barrera A thesis submitted to fulfil the requirements of the Department of Accounting and Finance of the University of Seville, Spain, to obtain the doctoral degree. June, 2017 GMB Doctoral Thesis 2/130 Doctoral Thesis: Real Options applications to assess the impact of public funding on R&D projects from a double perspective: the private company and the funding body. PhD Candidate: Gonzalo Martín-Barrera Tutors: Constancio Zamora-Ramírez & José María González-González Location: Department of Accounting and Finance, University of Seville, Spain GMB Doctoral Thesis 3/130 To Abengoa Research for allowing me to prove my beliefs. GMB Doctoral Thesis 4/130 Preface This doctoral thesis is born as a response to a challenge posed in my previous company, Abengoa Research, in which I was definitely decided to prove the valuation of Research and Development (R&D) Projects should be assessed with Real Option Valuation (ROV) and not with the well-know and widely spread Net Present Value (NPV). Before reaching there, I have an Engineering background and experience. However, in 2010 I decided to start a MBA in Copenhagen finding Corporate Finance is my passion. The feel of quantifying money, time and even actions was a complete new area of curiosity and investigation. After joining Abengoa Research in 2014, I proposed to Manuel Doblaré, CEO of such company, to undertake this doctoral thesis to prove the company we should think out of the box when assessing R&D. It has taken me about three years and a half to do all the investigation, set my new mindset, prepare articles, have enriching discussions with my tutors, many and many hours of English editing and great doses of patience to wait journals’ response. Finally, I am here today to present my doctoral thesis as my last contribution to R&D. I have joined a Yieldco company in which I have extrapolated Corporate Finance and Engineering to a next level. I would like to thank my tutors their availability and flexibility to change their initial mindset to a new paradigm in the Engineering. GMB Doctoral Thesis 5/130 Index Preface ........................................................................................................................................ 4 Index ........................................................................................................................................... 5 List of tables ............................................................................................................................... 7 List of figures ............................................................................................................................. 8 Acronyms ................................................................................................................................... 9 Introduction ............................................................................................................................. 10 Theoretical framework ........................................................................................................... 22 2.1. RO Theory .................................................................................................................. 23 2.2. Application of RO to R&D projects ............................................................................ 25 2.3. RO in policy evaluation of renewable energy projects ............................................. 27 2.4. Rationale for R&D grants: motivation, limitations and effect on companies ........... 29 2.4.1. Reasons to subsidise R&D investments ............................................................ 29 2.4.2. Limiting factors .................................................................................................. 30 2.4.3. Effect on companies .......................................................................................... 35 Methodology ........................................................................................................................... 37 3.1. Factors considered in R&D project valuation & Stage-gate framework ................... 38 3.2. Common R&D incentives in Spain and Europe ......................................................... 40 3.3. Qualitative effect of grants in terms of managerial flexibility .................................. 41 3.4. Case approach ........................................................................................................... 42 3.4.1. NPV calculation ................................................................................................. 44 3.4.2. Effect of grants in NPV ...................................................................................... 44 3.4.3. RO model ........................................................................................................... 44 First case: The Private Company ........................................................................................... 46 4.1. Project description .................................................................................................... 47 4.2. Parameter estimation ............................................................................................... 49 4.3. Scenario analysis ....................................................................................................... 54 4.4. RO Model................................................................................................................... 57 4.5. Varying the conditions .............................................................................................. 61 4.6 Discussion ........................................................................................................................ 61 GMB Doctoral Thesis 6/130 Second case: The Public Institution ...................................................................................... 64 5.1 Project description .......................................................................................................... 66 5.2 Parameter estimation ..................................................................................................... 67 5.3 Scenario analysis ............................................................................................................. 72 5.3.1 Original investment in 1998 ..................................................................................... 72 5.3.2. What would be the required grant to bring the project forward to 1996? ............ 72 5.4 RO model ......................................................................................................................... 74 5.5 Sensitivity analysis ........................................................................................................... 79 5.5.1 Sensitivity analysis: discount rate ............................................................................ 79 5.5.2 Sensitivity analysis: volatility .................................................................................... 80 5.3.2 What if the first commercial plant were built sooner? ............................................ 81 5.6 Combining parameters .................................................................................................... 82 5.7 Discussion ........................................................................................................................ 83 Conclusions .............................................................................................................................. 85 References ................................................................................................................................ 91 Appendix I .............................................................................................................................. 101 Appendix II ............................................................................................................................. 111 Appendix III ............................................................................................................................ 123 GMB Doctoral Thesis 7/130 List of tables Table 1: Eurotrough I & II start and end dates, declared costs and EU contribution .... 48 Table 2: NPV of commercial revenues discounted at indicated year .................................. 51 Table 3: Parameters for the binomial tree ...................................................................................... 52 Table 4: Base case cash-flow. All figures in k€. ............................................................................. 55 Table 5: Grant case cash-flow. All figures in k€. ........................................................................... 56 Table 6: Evolution of the underlying asset and real option value at each time step ..... 59 Table 7: Project value for each case: DCF with and without grant and ROA .................... 62 Table 8: Eurotrough I & II Start and End Dates, Declared Costs and EU Contribution . 67 Table 9: NPV of commercial revenues using Eurotrough rather than LS3 discounted at indicated year ............................................................................................................................................ 69 Table 10: Value of the parameters for the binomial tree of the original case (investment in 1998) ........................................................................................................................................................ 70 Table 11: Evolution of the underlying asset and real option value at each time step .. 76 Table 12: Sensitivity analysis of the Discount Rate ..................................................................... 80 Table 13: Sensitivity analysis of the Volatility................................................................................ 81 Table 14: Comparison of three different scenarios ..................................................................... 82 Table 15: SJR of RSER since 2006 .................................................................................................... 126 GMB Doctoral Thesis 8/130 List of figures Figure 1: ET150 versus LS-3 NPV improvement as function of storage hours ................. 50 Figure 2: Evolution of the underlying asset and real option value at each node ............ 58 Figure 3: NPV as function of the received grant .......................................................................... 61 Figure 4: Impact Factor of the journal ........................................................................................... 124 Figure 5: Ranking of the journal ....................................................................................................... 124 Figure 6: 5-year impact factor of the journal .............................................................................. 125 Figure 7: 5-year ranking of the journal .......................................................................................... 125 Figure 8: RSER SJR since 1999 ........................................................................................................... 126 Figure 9: RSER in Q1 since 2002....................................................................................................... 127 Figure 10: Google Scholar metrics .................................................................................................. 127 Figure 11: RSER is ranked #1 under Sustainable Energy by Google Scholar according to h5 index ..................................................................................................................................................... 127 Figure 12: RSER is ranked #11 under Engineering and Computer Science by Google Scholar according to h5 index .......................................................................................................... 128 Figure 13: RSER is ranked in the 96th percentile of Renewable Energy, Sustainability and the Environment according to CiteScore, being ranked #5 out of 117 journals. .......... 128 Figure 14: Citations received on Part 1 of this doctoral thesis ............................................. 129 Figure 15: First citation received, published in RSER in November 2016 ......................... 129 Figure 16: Second citation received, to be published in Renewable Energy in August 2017 ............................................................................................................................................................ 130 GMB Doctoral Thesis 9/130 Acronyms Concentrated Solar Power CSP Discounted Cash Flow DCF European Commission EC Feed-in Tariff FiT International Energy Agency IEA Megawatts electric MWe National Renewable Energy Laboratory NREL Net Present Value NPV Photovoltaics PV Real Option RO Real Option Analysis ROA Real Option Valuation ROV Renewable Portfolio Standard RPS Renewable and Sustainable Energy Reviews RSER Research & Development R&D Return on investment ROI Small and Medium-sized Enterprises SMEs GMB Doctoral Thesis 16/130 the application of NPV traditional method, then the project is valuated with RO methodology, in order to assess the loss of flexibility associated with the fact of accepting a grant. Specifically, the methodology employed consisted of the use of ROV to asses, at the end of each year, the economic sense it would have to continue with the R&D, to postpone it one year or even to abandon it. Companies have realized that R&D project management differs from traditional projects. The main uncertainty in the latter is the market risk that might affect future cash-in. However, an R&D project is affected not only by market evolution, but also by its technical success and the potential emergence of substitutes during the development of the product. Therefore, R&D requires a specific management framework to handle its risks and decisions to be taken during its lifecycle. In this sense, a new approach was developed to manage R&D projects, known as stage-gate methodology. This framework allows companies not to commit all the resources for a project at the very beginning, but it establishes several gates the project would pass or not depending on the technical success and the market evolution. If a project seems to be viable and the market perspectives are reasonable good, according to stage-gate methodology, the company should release more funds to continue the development. This tool ensures the company does not devote too many resources to a project about to fail or which will not be profitable in the future. Thus, the managerial flexibility is quantified (ie. the difference between the traditional NPV technique and the ROV). Grants not balancing the loss of flexibility are clearly a value destroyer. The specific example of this article presents a new solar tracker in 1998 with a subsidy granted by the European Commission; imposing deadlines to its finalization and not allowing the abandon option. With those limitations, a minimum grant of 17.25% is required, much lower than the 50% provided by the EC for that project. The question here is if the public institution excessively funded the project. GMB Doctoral Thesis 17/130 In order to complement the ROV from the private company perspective, this doctoral thesis considers also the public institution perspective. In this sense, private R&D investment in renewable energy technologies is often associated with market failure from the economic perspective (Gillingham & Sweeney 2010). This designation is employed to describe a situation in which there are no conditions to reach an efficient allocation of resources (Stiglitz 1988). Thus, companies that invest in this type of project do not realise the full potential benefits that said projects could generate. Specifically, these projects positively impact society as society benefits from a less polluted environment, for instance. In fact, society demands greater investment by private companies in renewable energy R&D projects that would naturally be carried out by companies under perfect market conditions. One of the main reasons justifying the use of public funding for private R&D investments is precisely to overcome this market inefficiency in order to reach the socially optimal level of development (Stiglitz 1988). In this way, public funding can help companies to receive some of the potential benefits that are generated by their R&D projects, as well as encouraging R&D projects that, due to their high cost and risk, would not be undertaken by companies despite being socially valuable projects (Hall 2002). When public funding for private R&D investment is considered, specifically by means of public subsidies, one of the questions that has attracted the attention of researchers regards the effect of these public subsidies on private R&D investment (David et al. 2000; Ali-Yrkkö 2004; Afcha & León López 2014). In particular, the specialized literature has distinguished two types of effect: the crowding-in effect, which implies that public subsidies to R&D projects tend to stimulate private R&D investment and therefore have an additive effect; and the crowding-out effect, which, on the contrary, implies that public subsidies to R&D GMB Doctoral Thesis 18/130 investments tend to reduce private investment with the result that public funding would negatively displace private R&D investment. More specifically, the main drivers for public institutions are the widest promotion of private R&D investments using the lowest public funding. There is a risk, the so-called crowding-out effect, by which above a threshold, more public funding reduces the private investment -ie. the public institution has granted excessively. This section fully explains to public institution how a grant could be equally attractive for companies by reducing the nominal grant but enhancing the flexibility embedded (at no cost for the Institution). As shown by Zúñiga et al. (Zúñiga-Vicente et al. 2014), the empirical evidence of this effect is inconclusive: they found differences that may be due to factors such as the countries and industries considered, or the empirical approach conducted. In their wide literature review, these authors highlight that most studies have not considered the amount of the subsidy granted to the firm in their empirical assessments. Various factors, however, support the existence of a level of inflection in the subsidy amount that reduces private R&D investment (for instance, the rising of specialized workforce costs (Wolff & Reinthaler 2008; Mohnen et al. 2008); the reduction of inflection in the efficiency of these subsidies for larger projects (Aschhoff 2009; Irwin & Klenow 1996; Setter & Tishler 2005; David et al. 2000); or that these projects stop having the desired level of efficiency in relation to the social benefits that justify the existence of these subsidies). Thus, the second goal of this doctoral thesis is, by considering the public institution perspective, to fill the gap in the existing literature by establishing a reference for policy makers and funding bodies to determine the maximum amount of public subsidies to be given to private R&D investment in order to accelerate companies’ R&D activity while reducing the crowding-out effect on private investment. In order to do so, this paper applies the ROA to the case of GMB Doctoral Thesis 19/130 an R&D project in solar thermal power carried out by a consortium of European companies. Similar to the analysis carried out from the private company perspective, a Real Options Analysis is performed to determine the value of flexibility such public institutions could provide (such as longer periods). By taking into account this added value, the nominal grant can be reduced and therefore the crowding-out effect risk decreases whereas the public institution has more resources to grant additional projects and companies. Furthermore, the numerical case applied to illustrate this rationale is the same as the case employed in the analysis from the private company perspective; then all the background is shared. However, as in the private company perspective the focus was set on the company and how much should be the minimum grant to accept the project; in the public institution perspective, the focus is switched to the public institution and which alternative it has to promote an equally attractive project but minimizing the risk of crowding-out (ie. decreasing the private R&D investment). The use of the same case for both perspectives is intentional to further enrich the work by providing the two points of view of the same numerical example. Traditionally, public subsidies to R&D programs, such as the European Commission’s Horizon2020 program (EuopeanCommision 2014), restrict the flexibility of companies when deciding whether to continue their project. We did not confirm the anticipated results in this study; the decision to abandon a project, if this were the case, would imply the partial repayment of the public subsidy – if the funding body considers the project has not reached a minimum level - which, in turn, would reduce the probability of being granted future subsidies as a consequence of the reputation loss of the company towards the funding body. Likewise, since the probability of a crowding-out effect can be expected to increase with the subsidy size (Siddiqui et al. 2007), other things being equal, the proposal of this paper also contributes to reducing the GMB Doctoral Thesis 20/130 probability of said effect because it justifies the reduction of the public subsidy to private R&D investment. Hence in restructuring classical grant schemes by giving firms options, these companies could decide to undertake projects a few years earlier and accept lower levels of grant funding. Thus, a more flexible scheme could reduce the crowding-out effect, deferring the inflection point where grants are less efficient for attracting private investors. Therefore, the main contributions of this doctoral thesis are, firstly, to assess the loss of flexibility imposed to management, and therefore the resulting need to determine the lower limit of a grant from which companies should not accept public R&D funds. Previous studies about R&D projects in renewable energy valuation have not considered the loss of flexibility imposed by grants (Santos et al. 2014; Martínez-Ceseña et al. 2013; Tsui 2005; Hodota 2006; Trang et al. 2002). In this respect, this doctoral thesis contributes to strength the RO framework to assess energy R&D projects and so, allowing investors to be more conscious about costs involved in government grants. Also, ROA has been used to policy evaluation (Siddiqui et al. 2007; Yu et al. 2006; Boomsma et al. 2012; Lee et al 2011; Lin et al. 2013; Reuter et al. 2012; Monjas-Barroso et al. 2013; Zhang et al. 2014). In this sense, policy makers will have a new point of view when defining public R&D: the jeopardy of managerial flexibility. A very restrictive grant policy would negative impact in the development of projects whereas allowing some actions to the companies would increase the projects’ value without representing an extra cost to governments. Secondly, this doctoral thesis contributes to fill the gap in the existing literature by establishing a reference for policy makers and funding bodies to determine the maximum amount from public subsidies to be given to private R&D investment in order to accelerate companies’ R&D activity while reducing the crowding-out effect on private investment. In particular, this doctoral thesis contributes to the specialized literature by proposing an assessment system GMB Doctoral Thesis 21/130 based on ROA that justifies the reduction of the amount of public subsidies for private R&D investment by showing that companies could prefer a lower public subsidy for their R&D investment, but that the public subsidy allows them greater flexibility by permitting options such as abandoning or deferring the R&D project. The reminder of this doctoral thesis is structured as follows. The second chapter presents, on the one hand, a review of the literature regarding RO and specifically applied to R&D projects and, on the other hand, and in a complementary way, a review of literature related to the main factors that can impact the relationship between public subsidies to R&D and private R&D investment, as well as the specific influence of the subsidy size on the R&D activity of private companies. The third chapter describes the methodology employed in this work in order to reach our two goals, whereas the fourth and fifth chapters present, respectively, the results of the empirical studies carried out, as well as the discussion of the results. Finally, the sixth chapter presents the main conclusions of this doctoral thesis, indicating limitations to the current research and further research to this work. GMB Doctoral Thesis 22/130 Chapter II. Theoretical framework GMB Doctoral Thesis 23/130 This chapter presents the theoretical framework of this doctoral thesis. Firstly, a review of literature regarding RO is presented, by considering in particular the evolution of the RO theory, its application to R&D projects, and its consideration for the policy evaluation of renewable energy projects. Secondly, the previous literature review is complemented with the rationale for R&D grants, by considering specifically the reasons to subsidize R&D projects, the main factors that can impact the relationship between public subsidies to R&D and private R&D investment, as well as the main effects on companies. 2.1. RO Theory A financial option is the right but not the obligation to trade products at a specific time for a predetermined price (Ross 2003). These products are contracts between two parties, typically consumers and sellers. RO is the extension of that financial theory but for real assets assessments. Therefore, a RO can be defined as the right but not the obligation to make an investment decision concerning real assets (Menegaki 2008), for instance defer its development, build a prototype, abandon the project, alter its scale, switch to another technology, etc (Martínez-Ceseña et al. 2011). These possible actions are known as managerial flexibility, and it enhances the value of the projects. There are two options studied in this paper: one is the defer option, giving the manager the ability to wait for investing the money. The value of this time of waiting comes as a result of the more information about the investment obtained or the resolution of an uncertainty (such a new law or regulation in process) (Dixit et al. 1994). The second one is the abandon option, which reflects in a more realistic manner the reality of the project; if the evolution of the market or the technical success is not as expected, the company can stop investing in a non-profitable project. GMB Doctoral Thesis 24/130 In other words, RO account for management flexibility which delivers a significant value contribution in the presence of uncertainty. Technically, the RO method is more appropriate for valuation of R&D projects than DCF, because plans are options, not assets. For a real option, the underlying security is the present value of the business plan (Boer 2004). Special features of energy R&D projects suggest ROA to be more suitable than traditional methods like NPV to assess their economic valuation, as shown by Santos et al. (Santos et al. 2014).In most cases, energy R&D projects are irreversible because associated investments are partially or completely irreversible. Therefore, the assets of these projects cannot be used in other activities or different companies. Besides that, there is a high level of commercial uncertainty, especially in the liberalized electricity market (Kumbaroğlu et al. 2008). Moreover, investment can occur in a flexible time. Finally, by designing the project, investors have several generation technologies that are associated with different uncertainty levels. Therefore, ROA has been recommended to be more appropriate than traditional NPV for judging R&D projects (Hartmann et al. 2006). Myers (Myers 1977) was the first who recognized the analogy between financial options and real world investments, being the one who coined the expression “real option”, based on the resemblance between R&D investments can be valued similar to financial options. Mapping an investment opportunity onto a call option (Olilla 2000) implies the stock price would be equal to the present value of the project’s operating assets to be acquired (or developed); the exercise price is the expenditure required (or the R&D investment), the time to expiration would be the length of time the decision may be deferred, the risk-free rate of return equals the time value of money and finally the variance of returns on stock, the riskiness of the projects assets. The evolution of financial options began in the early 70s due to growing dissatisfaction with DCF methods at the same time significant breakthroughs in GMB Doctoral Thesis 25/130 options theory were developed by Fisher Black, Myron Scholes (Black et al. 1973) and Robert Merton (Merton 1973) who were the first to price their value. Later on, Cox introduced the pricing formula for binomial trees (Cox et al. 1979). Decision trees only capture the substantial value of the option to abandon; however, RO captures the value from the management of market risk (risk that cannot be diversified) who adapt the project to the market circumstances. However, other approaches have been taken for getting the valuation of RO. For instance, Amran and Kulatilaka (Amram et al. 1999) use partial differential equations to express mathematically the value of an option and its dynamics (Fernandez et al. 2011). Finally, another approach is to simulate the uncertainty with Monte-Carlo. 2.2. Application of RO to R&D projects The problem with DCF is the mind-set that has developed around it. For example, the Japanese approach to R&D valuation is not consistent with DCF thinking, because growth and market are more important than return on investment (ROI) (Faulkner 1996). Despite of that, DCF is still commonly used in R&D valuation. However, there are several cases in literature in which RO are applied to R&D projects. For instance, Tsui (Tsui 2005) applies the RO theory to fuel cell vehicles, Hodota (Hodota 2006) to an intelligent transportation system, and Trang (Trang et al. 2002) to the pharmaceutical industry. All of them valuate the project with DCF that implies no flexibility. Over that value, normally a decision tree is considered, increasing the value due to the abandon option. Finally, RO provide the higher value because it accounts for all the potential flexibility related to the project. In the field of renewable energies, Venetsanos et al. (Venetsanos et al. 2002) proposed a framework for the appraisal of power projects under uncertainty GMB Doctoral Thesis 32/130 2.4.2.3. Economic context In relation to small and medium-sized enterprises but also considering the influence of the economic context in their study, Hud & Hussinger (Hud & Hussinger 2015) analysed the impact of public R&D subsidies on the R&D investment of German companies during the latest economic crisis. They found an overall positive effect of R&D subsidies on R&D investment attitude from companies, although there is evidence of a crowding-out effect during the 2009 crisis. Furthermore, when the German economy started to recover in 2010, the subsidy effect was smaller than in the pre-crisis years. Wiser & Pickle (Wiser & Pickle 1998) also considered the influence of the economic and political context in which companies develop the power plant financing process for renewable energy projects. They conducted five case studies showing policies that do not provide long-term stability or that have negative secondary impacts on investment decisions will increase financing costs and therefore could dramatically reduce policy effectiveness. This work specifically highlights that the importance of policy stability to renewable energy developers and financial investors should not be underestimated. In fact, changes in renewable energy subsidies have tended to be abrupt and therefore disruptive to developers and investors. 2.4.2.4. Type of incentive The effect of the type of incentive on the relationship between public R&D subsidies and private R&D investment has also been considered by specialised literature. In this vein, Guellec & Potterie (Guellec 1997) analysed whether tax incentives and direct subsidies stimulate business-funded R&D in 17 OECD countries. They found that although tax incentives and direct subsidies stimulate GMB Doctoral Thesis 33/130 private R&D investments in the short run, direct subsidies are more effective than tax incentives in the long term. In a later study also considering figures from 17 OECD countries, Guellec & Potterie (Guellec & Van Pottelsberghe De La Potterie 2003) concluded that both direct government funding of R&D performed by firms and R&D tax incentives have a positive effect on business financed R&D and that they are more effective when they are stable over time. Another important conclusion of Guellec & Potterie (Guellec & Van Pottelsberghe De La Potterie 2003) was that direct government funding and R&D tax incentives are substitutes. Thus, the increased intensity of one reduces the effect of the other on private R&D. More recently, Abolhosseini & Heshmati (Abolhosseini & Heshmati 2014) compared feed-in-tariffs, tax incentives and tradable green certificates as the main mechanisms used by governments to finance renewable energy development programmes. They concluded that tax credits constitute an attractive mechanism for private investors because they directly increase investor liquidity. On the other hand, while feed-in-tariffs are a suitable mechanism as a policy to develop renewable energy sources when low investor risk is preferred, tradable green certificates constitute the most appropriate policy when the government applies a market view policy. 2.4.2.5. Recurrence in public funding Regarding the recurrence of being granted public funding, the literature indicates that a firm whose R&D activity was subsidised in the past is more likely to be subsidised again (González & Pazó 2008). Nevertheless, Zúñiga et al. (Zúñiga-Vicente et al. 2014) noted that there are different reasons supporting alternative assumptions about the effect of public subsidies on private R&D investment. For example, because firms that are granted public subsidies for their R&D are more likely to benefit from grants that reduce their own risk and cost of financing R&D projects, the odds of a crowding-out effect for frequent GMB Doctoral Thesis 34/130 recipients of R&D subsidies are increased (Lööf & Heshmati 2005). Alternatively, it can be argued that because firms that have received a subsidy for a R&D project may subsequently apply for new subsidies to keep financing the same project, it can be expected that public R&D subsidies have a crowding-in effect on these firms’ R&D investments (Duguet 2003). 2.4.2.6. Size of the subsidy granted to firms One of the most debated issues has been the effect of the grant size on the crowding-in and crowding-out effects; in other words, does increasing the amount of the subsidy also increase private spending invested in R&D projects, or conversely, does it lead to decreasing R&D expenditures by firms? Several arguments have been provided by the literature in order to explain why R&D subsidies can lead to the displacement of private R&D. There are contrary reasons for supporting “the more the better”, resulting in the crowding-out effect of private funding. This means public funding may result in an undesirable replacement effect of the private investment. The main factor hindering R&D activities is the lack of qualified personnel (Mohnen et al. 2008). An inefficiency, as a result of R&D activities supported by public grants, is the higher demand for R&D personnel, increasing their wages (Wolff & Reinthaler 2008). R&D subsidies can easily translate into researcher wage increases (Goolsbee 1998). The size of projects can be another variable to be considered as a cause of displacement of private R&D investment (Aschhoff 2009). First, stimulation effects are more likely for large projects than for small ones. Thus, small projects have a higher tendency to lead to a crowding-out effect than larger projects. Second, a firm’s R&D capacity cannot be extended at will in the short-run even if the firm is awarded public support for a project. Firms cannot be willing to GMB Doctoral Thesis 35/130 make excessive long term employment commitments. If larger projects are subsidised and considering a restricted R&D employment market, firms undertake the large projects, reallocating funds and employers from other projects to the subsidised project and abandoning those previous projects. This would result in a reduction of private R&D investment. Other dangers of excessive grants are financing duplicate R&D and redirecting financing towards more risky (Irwin & Klenow 1996) and thus less successful projects (Setter & Tishler 2005). Moreover, government incentives can lead to the creation of expensive R&D facilities (David et al. 2000), but firms equipped with this R&D infrastructures may continue to apply for new subsidised R&D projects in order to cover these fixed costs, without regard for improved economic performance and the advancement of knowledge. The project could be concluded without social benefits and the firm would not profit beyond the coverage of its fixed costs. These firms with more experience and better equipment for R&D could prevent other companies with better projects from being awarded R&D subsidies. 2.4.3. Effect on companies As Aschoff (Aschhoff 2009) proposes, there are different effects of grants on a firm’s R&D expenditures: a full crowding-out effect, if R&D subsidies substitute company expenditures; partial crowding-out, if the firm’s expenditures in R&D would decrease in the situation of public funding; no effect of subsidies on a firm’s own R&D; and an additional effect, if the firm increases privately financed R&D spending when subsidies are received. The literature shows ambiguous results regarding these findings. A full crowdingout effect has been rejected by (Aerts et al. 2006; Clausen 2007; Ali-Yrkkö 2004; Kaiser 2006; Heijs & Herrera 2004; Görg & Strobl 2007), whereas Lach (Lach 2002) GMB Doctoral Thesis 36/130 found a partial replacement of private R&D expenditures with public funding. Streicher et al. (Streicher et al. 2004) found that private funding increases when R&D projects are subsidised by public grants. Aerts (Aerts 2008) asserted these mixed results are due to the heterogeneity of samples belonging to different countries and industrial sectors. Görg and Strobl (Görg & Strobl 2007) found additional effects of small grants, rejecting the crowding-out of medium grants. Aschhoff (Aschhoff 2009) also showed that grants should have a minimum size to cause an impact on a firm’s privately financed R&D and that medium- and large-scale grants increase firms’ R&D spending depending on the size of the project. Larger projects are more dependent on public funding. Guellec and Pottelsberghe (Guellec & Van Pottelsberghe De La Potterie 2003) found an inverted U-shaped curve in the relationship between public subsidies and private expenditures in R&D funding. The subsidy effect is positive but marginally decreasing up to a certain threshold, beyond which the effect becomes negative. This confirms a crowding-in effect of moderate grants and a crowding-out effect for subsidies beyond a certain level (Zúñiga-Vicente et al. 2014). GMB Doctoral Thesis 37/130 Chapter III. Methodology GMB Doctoral Thesis 38/130 This section of the doctoral thesis explains the methodology utilized to determine the value of the flexibility for both the companies and the public institutions for specific R&D projects. 3.1. Factors considered in R&D project valuation & Stage-gate framework Companies have realized that R&D project management differs from traditional projects. The main uncertainty in the latter is the market risk that might affect future cash-in. However, an R&D project is affected not only by market evolution, but also by its technical success and the potential emergence of substitutes during the development of the product. Therefore, R&D requires a specific management framework to handle its risks and decisions to be taken during its lifecycle. In this sense, a new approach was developed to manage R&D projects, known as stage-gate methodology (Cooper et al. 2018). R&D projects, in contrast with traditional large projects, differ mainly in the high uncertainty related to the investment, not only due to the uncertainty of the market demand for the product being developed but also to the probability of technical success. Therefore, companies must adapt their investment in order not to commit excessive resources to a potential failure. Cooper (Cooper et al. 2018) proposed a scheme by which each phase of the R&D project – namely the initial design, advanced design, first prototype, scale-up, and commercial - is given a specific budget and deadline. By doing so, the project is forced through a review process before beginning a new phase. These periodic revisions allow the managers to pause the development, cancel (kill) the project if the expectations are not promising, or the reverse, boost investment to reach the goals earlier than originally planned. These options on the part of managers are essential to the point of this article. First, the mere option of companies to abandon the project adds value to the R&D investment because not all the resources must be committed at the outset. GMB Doctoral Thesis 39/130 This may seem obvious; however, using traditional DCF techniques, this possibility is not modelled. Second, the company has the option to defer (pause) the project for a few years to gain information about the market or the technology. This delay also has value because if the new market expectations are more promising, the company can continue the R&D project while retaining the possibility to abandon it in the future. Again, traditional DCF cannot include this deferment action in the model. Third, if the technical success seems to be even better than initially planned or the market demand is greater, the company can allocate more resources to finish the R&D project earlier and therefore arrive sooner to the market. As DCF does not model any change during simulations, this possibility is again neglected. This framework allows companies not to commit all the resources for a project at the very beginning, but it establishes several gates the project would pass or not depending on the technical success and the market evolution. If a project seems to be viable and the market perspectives are reasonable good, according to stage-gate methodology, the company should release more funds to continue the development. This tool ensures the company does not devote too many resources to a project about to fail or which will not be profitable in the future. As mention above, lying behind a stage-gate decision, there are at least two RO. One of them is the option to abandon the project (at each gate) and the second would be the decision to defer the project. For instance, it could be a defer decision based on the current market estimations, or due to a certain regulation under discussion. The project could be delayed one year and re-evaluated to decide then either to continue or not with more information available. To evaluate an R&D project using NPV, the first step is to estimate the total market for the technology and the correspondent market share. This market share would be translated into a cash-in per year during the commercialization phase. Prior to that, an R&D investment phase is necessary in which there is cash-out GMB Doctoral Thesis 40/130 from the company. Using discounted cash flow, the future cash-in would be discounted together with the cash-out. If the NPV is above cero, the project would be undertaken. The main issue applying DCF is to determine the discount rate, which is related to the cost of capital of the company, representing its opportunity cost. The second uncertainty would be the technical success of the R&D phase (Siddiqui et al. 2007), in which NPV “simulates” private risk by adding subjective probabilities. Using Monte-Carlo simulation, a probability distribution of the NPV can be obtained, depending on the market and technical variables and probabilities associated to each iteration. In the analysis of R&D projects using DCF, the effect of grants is always beneficial. Grants will reduce the R&D investment phase; therefore the NPV will increase no matter the conditions imposed to the project by the funding agency, because DCF cannot handle them and they are not included in the numerical analysis. 3.2. Common R&D incentives in Spain and Europe There are plenty of incentives for R&D projects, as explained in the literature review of section 2. The subsidies most frequently used in Spain and given by the European Commission are direct funds to undertake an R&D project. Tax incentives are more oriented to past R&D activities or the commercial exploitation of results rather than to initiate new projects. Depending on the technical maturity of the technology, being the distinction between Research and Development, and the type of the beneficiary of the incentive (privately or publicly held, small, medium or large company), there is a range of incentives between 25% and 100%. For mature projects, Spanish funding starts at 25% for large companies. This funding rate increases up to 100% for Research projects and/or R&D centres. GMB Doctoral Thesis 41/130 In addition to the financial help, there is a drawback for companies accepting these incentives, namely, the lack of flexibility. Accepting one of these grants imposes a deadline to finish the project based on a project chart and the minimum execution of budgeted costs. The three options explained above contrast somewhat with the traditional way of valuing incentives. These drawbacks are completely ignored by DCF because this model would not include the lack of flexibility imposed nor the potential advantage of developing a technology or product earlier or faster. Using the RO framework, a more suitable tool for valuing these options will be taken into account and a recommendation for funding bodies will therefore be proposed. 3.3. Qualitative effect of grants in terms of managerial flexibility Real option theory is well known for enhancing the value of projects under uncertainty (Martínez-Ceseña et al. 2013). In R&D projects this managerial flexibility would be only considered if managers are willing to exercise the options. This might not be always the case if managers cannot or are reluctant to apply all the options, for instance they could be not willing to abandon their project. In those cases the use of RO is useless because it provides a value considering a managerial flexibility that is not real. As stated before, stage-gate is a well-known methodology to manage R&D projects in which there are at least two RO embedded: the option to abandon the project at each gate and the option to defer the investment. Scholars have deeply analysed how to apply RO to a project model to include the managerial flexibility. Particularly, Fernandes et al. (Fernandez et al. 2011) analyse the ability to delay an investment, in order to obtain more information and thus reducing uncertainty. GMB Doctoral Thesis 48/130 due to a new design, reduction of operation and maintenance costs due to easier mirror cleaning, less complex realignment requirements, and fewer components. The project name was ‘Eurotrough’ and it was made by a consortium of companies in the late 90s (Ciemat et al. 2014). The project received two grants from the European Commission (EC) under two different contracts covering sequential stages of the development. Table 1 summarizes the total declared costs and public R&D funds from the European Union (EU contribution) per contract. Table 1: Eurotrough I & II start and end dates, declared costs and EU contribution Start date End date Total declared costs EU Contribution Eurotrough I 1-Ago-98 31-Jan-01 2.402.514 € 1.199.899 € Eurotrough II 1-Oct-00 31-Dec-02 1.993.707 € 996.851 € Even though CSP plants are complex, large installations with hundreds of design parameters (European Solar Thermal Electricity Association 2014), there are two particularly important that roughly determine the size and cost: electrical capacity which defines the size of the steam turbine; and the storage capacity expressed as how many hours the plant is able to work at full load with no sun. There is a positive correlation between those two with the size and therefore the cost of the plant. The larger the plant is, the more number of solar reflectors are required and therefore the improvement per tracker, due to having more efficient troughs, will be more substantial. There is a limit in this correlation whereas adding more trackers do not improve the profitability of the plant. GMB Doctoral Thesis 49/130 4.2. Parameter estimation The first step is to calculate the improvement the new trough brings over the previous one. To do so, a tool, called System Advisor Model, designed by the National Renewable Energy Laboratory (NREL) from United States is run (NREL 2014). This tool enables to simulate the behaviour of all type of CSP plant in any location worldwide, defining its main parameters such as electrical capacity, storage size and, among others, model of trough. The output is the NPV of the designed plant. Simulations 1 were run with the trough previous to the R&D project (LS-3) and with the developed one (ET150). For both troughs, 10 locations were simulated 2 with different storage hours 3 covering all the normal ranges of design. Therefore, for each number of storage hours, the average (of the 10 locations) of the difference in both troughs’ NPV was calculated, with the results provided in figure 1. The figure is normalized per unit of installed MW. 1 Parameters of the simulation: Parabolic Trough physical model, Utility Independent Power Producer, electrical output of 50 MW, 4 troughs per loop, tracking error of 1.00 for LS-3 and 0.99 for ET150, the remaining parameters as default except the model of 2 types of troughs (LS-3 and ET150), 10 locations and 16 storage hours. 2 Phoenix, Tucson, Colorado Springs, Grand Junction, Albuquerque and Tonopah (US), El Cairo Int Airport (EGY), Almeria, Granada and Sevilla (ESP). 3 From 0 to 15 in steps of 1h. GMB Doctoral Thesis 50/130 Figure 1: ET150 versus LS-3 NPV improvement as function of storage hours These results allow to calculate market revenues of the technology: NREL (NREL 2014) has registered all the commercial CSP power plants worldwide indicating electrical capacity, type of trough (LS-3, ET150 or others) and storage size. Applying the function of improvement in NPV as function of the storage hours per installed MW to all the CSP plants retrieved from NREL, the worldwide NPV per year is calculated and shown in table 2. GMB Doctoral Thesis 51/130 Table 2: NPV of commercial revenues discounted at indicated year 2006 243 159.55 € 2007 952 284.55 € 2008 1 296 778.64 € 2009 1 398 113.19 € 2010 3 343 509.25 € 2011 2 703 466.25 € 2012 136 322.02 € 2013 1 013 243.07 € Total 11 086 876.50 € For the purpose of this paper, there are no more Eurotrough structures foreseen to be installed in any future CSP plant. The NPV is discounted to two years prior the plant starts operating, because the construction time is, on average, 24 months. The sum of these NPV discounted at 1998 (the year the R&D project was initiated) is the value of the underlying asset. GMB Doctoral Thesis 52/130 Table 3 summarises all the parameters that will be applied in the case: Table 3: Parameters for the binomial tree Stock price S0 3 634 194.92 € Discount rate r 10% Volatility σ 30% Risk-free rate rf 5% Exercise price 1 X1 808 838.00 € Exercise price 2 X2 808 838.00 € Exercise price 3 X3 808 838.00 € Exercise price 4 X4 996 853.50 € Exercise price 5 X5 996 853.50 € Duration stage 1 T1 1 Duration stage 2 T2 1 Duration stage 3 T3 1 Duration stage 4 T4 1 Duration stage 5 T5 1 Incremental time step δt 1 u = exp (σ√δt) 1.34986 d = 1/u 0.7408 p = (exp(rfδt)-d)/(u-d) 0.5097 (1) S0, as stated before, is the value of the underlying asset, calculated discounting to 1998 all the NPV of every single CSP plant using Eurotrough built worldwide. (2) Discount rate (r) is the opportunity cost of the company undertaking the investment. For confidentiality, no discount rate of any of the companies in the consortium has been used; but a figure (10%) in the range. Using another discount rate would change the absolute results, requiring a higher or lower grant, but it would not affect to the analysis nor the discussion. GMB Doctoral Thesis 53/130 (3) Risk-free rate (rf) is the discount rate assumed to be riskless discount rate, on the basis of the U.S Treasury spot rate of return, with is maturity equivalent’s to the option’s time to maturity. (4) Volatility (σ) is the risk of the expected returns. The market risk has been eliminated by using a technology already implemented. However, there is a risk associated to the performance of the CSP plant, depending on the weather, the availability factor of the plant, variations in the price of labour or some commodities such as water for cleaning mirrors, etc. After simulations were run, a deviation of 30% was detected. Identically to the discount rate, a different value would require a different minimum grant, remaining the validity of the discussion. (5) Exercise price (of period) i (Xi) means the cost of undertaking the following stage. The project has been divided into 5 stages (being one year-long each of those) in which the company can decide whether to continue (exercising the option at price Xi) or abandon at current stage. The cost of each period is extracted from table, assuming Eurotrough 1 is undertaken in 1998, 1999 and 2000 (being the same investment per year, although the period in 1998 is shorter) and Eurotrough 2 into 2001 and 2002. (6) Duration stage (Ti) is the length of each stage. It is assumed to be one year each, representing the company could abandon the project at the beginning of each new year. (7) Incremental step (δt) is each one of the discrete periods the model is divided. (8) u and d : are the RO parameters applied to the binomial tree according to (Kodukula et al. 2006) and (Hull 2009). So is the initial value of the asset, but in the first time increment, it either goes up or down and from there continues to go either up or down in the following time increments. The up and down movements are represented by u and d factors, where u >1 and d =1/ u . The magnitude of these factors depends on the volatility of the underlying asset. GMB Doctoral Thesis 54/130 (9) p : is the risk-neutral probability that enables to uses the risk-free rate as discount rate. 4.3. Scenario analysis With the use of the previous parameters, the NPV at 1998 of the R&D investment without any grant is 0.01 M€, which means a ROI of roughly 0%. This cash-flow has considered the five R&D investments in 1998-2002, three idle years (2003, 2004 and 2005) and commercial cash-in (from table 2) in the period 2006-2013. The detailed cash-flow is in table 4. However, if the grant is considered, the previous NPV improves significantly. Using the grant given by the EC (50%) supposed to be in 1999 for Eurotrough I and 2002 for Eurotrough II –this assumption states the grant is received the following year after its concession; the NPV of the project increases up to 1.785 M€ (ROI equals to 41%). The detailed cash-flow is in table 5. In this scenario, there is no way of introducing the loss of flexibility since the DCF does not allow to valuate it GMB Doctoral Thesis 55/130 Table 4: Base case cash-flow. All figures in k€. 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Eurotrough I - 801 - 801 - 801 Eurotrough II - 997 - 997 Grant Commercial revenues 243 953 1 297 1 398 3 344 2 703 136 1 013 Total - 801 - 801 - 801 - 997 - 997 - - - 243 953 1 297 1 398 3 344 2 703 136 1 013 - discount rate: 10% NPV (@1998) 14 Total R&D investment - 4 396 ROI 0% GMB Doctoral Thesis 56/130 Table 5: Grant case cash-flow. All figures in k€. 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Eurotrough I - 801 - 801 - 801 Eurotrough II - 997 - 997 Grant 1 200 997 Commercial revenues 243 953 1 297 1 398 3 344 2 703 136 1 013 Total - 801 399 - 801 - 997 - - - - 243 953 1 297 1 398 3 344 2 703 136 1 013 - discount rate: 10% NPV (@1998) 1 785 Total R&D investment - 4 396 ROI 41% GMB Doctoral Thesis 57/130 4.4. RO Model Finally, applying RO (without grants) implies building a binomial tree with the following results showed in table 6. This table is divided in seven sections: the first area indicates the theoretical value of the underlying asset, starting with S0 which is the NPV of the commercial revenues at 1998 and in each node the value can either go up by u or down by d. The second section is the nominal value of the underlying asset. Thirdly, the option value of stage 5 calculated following (Hull 2009) by which final node equals to the maximum of the underlying asset value minus the exercise price or zero (if the exercise price is greater than the asset, the option value is zero because it is not executed). This last node does not account for the defer option, only for abandon since it is the last opportunity the company can exercise the option and therefore execute the R&D project. Therefore, the option value in the last node, for instance S0·u^6 would be max {S0·u^6– X5 ; 0}. All the terminal nodes are calculated the same way. Going backwards to intermediate nodes, at each step there is a comparison between exercising the option at that node and waiting for the next time step. In this case, for example the valuation of the node S0·u^5 would be [p·(S0·u^6) + (1-p)·(S0·u^5·d)] · exp (-r δt). Once all the option values for stage-5 have been calculated, in order to determine stage-4 option values, the same procedure applies but using as underlying asset the stage-5 option values. Once stage-4 is calculated, it is possible to compute stage-3 and so on until stage-1 is determined. Value of stage-1 at time zero is the real option value for the project. GMB Doctoral Thesis 64/130 Chapter V. Second case: The Public Institution GMB Doctoral Thesis 65/130 This case is a new perspective on the situation described in (Martín-Barrera et al. 2016), in which a company faced the dilemma of accepting or rejecting a grant to finance a R&D project. In that case, besides the financial benefit of the grant, the drawback was the reduction of flexibility imposed by the European Commission. Therefore, a minimum grant was required to compensate the value lost on flexibility. This new perspective raises a different approach, however: the recommendation will be aimed at the financing body rather than the companies. The purpose of this numerical case is to illustrate the maximum grant a financing body should offer to companies in order to accelerate their research, this value being lower than the 100% that may crowd out private investment. In other words, because there is a minimum grant above 0% companies should require, there must be a maximum grant below 100% financing bodies should not exceed. Although this case is based on a specific technology, the value of its presentation resides in the possibility of subsequently extrapolating the methodology to any other technology. The use of ROV is not Renewable Energy specific, but applicable to any R&D project undertaken by a private company or group of companies which presents flexibility. Furthermore, this case is not dependent on the European system of grants, but applies to any subsidy provided worldwide which provides an economic incentive to private companies while limiting flexibility in any way -setting a deadline, for example. In this section, we present the numerical case in which we compare the original investment the company made and the maximum equivalent grant given by a financing body (the European Commission in this case) that would have been necessary to accelerate the project by a couple of years. Subsequently, due to the GMB Doctoral Thesis 66/130 high sensitivity of certain parameters, a further analysis is undertaken under different assumptions. As previously mentioned, the choice of an old project is intentional for two main reasons: (a) The main difference in using an ongoing project would be the need to project the total market and the corresponding market share captured by the R&D project results. This would complicate the analysis by adding more uncertainty to the calculations and deviating attention from the ROV. The significant advantage of using old technology is that all the potential revenues of the project are perfectly defined. Furthermore, (b) this case was presented by Martín-Barrera et al. (Martín-Barrera et al. 2016) from the company’s point of view. By using the same case, this analysis complements the previous one by adding the funding body’s perspective. Thus the full analysis is performed and determines on the one hand the minimum grant a company should accept and on the other, the maximum grant a funding body should offer. 5.1 Project description This project describes the development of a new component for Concentrated Solar Energy (CSP). This technology complements the traditional Photovoltaics (PV) using the Sun as source of energy. Together, there could be more than 400 GW of electrical capacity installed by 2035, according to the International Energy Agency (IEA) (IEA 2008). CSP is divided into two technologies: tower and troughts; however, both are based on the same principle which is to concentrate sunlight using mirrors and produce heat to drive a steam turbine. The tower uses trackers that follow the sun and concentrate light onto a solar receiver on the top of the tower. The trough (or parabolic trough) consists of a linear parabolic reflector that concentrates light onto a receiver positioned along the reflector’s focal line (European Solar Thermal Electricity Association 2014). GMB Doctoral Thesis 67/130 This R&D project seeks the development of a new 150-metre-long trough as opposed to the former 100 metre LS-3 used by its competitors. Besides the economies of scale provided by a larger component, this tracker has a lighter design that reduces operation and maintenance costs due to easier mirror cleaning, and fewer components. The R&D project name was ‘Eurotrough’, and it was carried out by a consortium of companies in the late 1990s (Ciemat et al. 2014). Eurotrough received two different grants from the European Commission (EC) called Eurotrough I and Eurotrough II. The below table describes the start and end dates, the cost declared by the consortium, and the grant received from the EC called the EU contribution. Table 8: Eurotrough I & II Start and End Dates, Declared Costs and EU Contribution Contract Start Date End Date Total Declared Costs EU Contribution Eurotrough I 1-Ago-98 31-Jan-01 2 402 514 € 1 199 899 € Eurotrough II 1-Oct-00 31-Dec-02 1 993 707 € 996 851 € The project started in August 1998 with a grant of approximately 50% of the total declared cost. The question is: how much grant money would have been necessary to bring forward the investment by two years, i.e., to 1996? Of course, a 100% grant will always incentivise companies, but we argue that this is not the optimal funding rate. 5.2 Parameter estimation Although CSP plants are large, complex installations with hundreds of design parameters (European Solar Thermal Electricity Association 2014), there are two particularly important aspects that roughly determine the size and cost: the nominal output, which defines the size of the steam turbine, and the storage GMB Doctoral Thesis 68/130 capacity, expressed as the number of hours the plant is able to work at full load with no sun. The higher any of these two parameters is, the larger the solar field is. A larger solar field positively impacts on the benefits of using more efficient trackers since the benefit is multiplied more times over. The improvement of the new trough over the previous one is clearly described in (Martín-Barrera et al. 2016), in which different locations were simulated using incremental storage hours. The results are given in figure 1 of the previous chapter which shows the incremental NPV per hour of storage, normalised to electrical megawatt (MWe) of installed capacity. With these results, determining the market revenues of the technology is straightforward because all the commercial power plants that use this technology are already known and no new plants will be built using that technology. The fact of using an old R&D projects helps to determine the exact market revenues. If the project were online, these revenues should be estimation based on market share projections. NREL (NREL 2014) details all the commercial power plants using Eurotrough worldwide, combining the list (which details electrical capacity and number of storage hours) and the increments in NPV per MWe and storage hour given in figure 1. The total NPV due to technology at the beginning of each year is shown in the below table. GMB Doctoral Thesis 69/130 Table 9: NPV of commercial revenues using Eurotrough rather than LS3 discounted at indicated year Year Commercial revenue of R&D project 2006 243 159.55 € 2007 952 284.55 € 2008 1 296 778.64 € 2009 1 398 113.19 € 2010 3 343 509.25 € 2011 2 703 466.25 € 2012 136 322.02 € 2013 1 013 243.07 € Total 11 086 876.50 € to assess the value of flexibility embedded in the project, the necessary parameters for the use of ROV are defined in the following table and explained here. GMB Doctoral Thesis 70/130 Table 10: Value of the parameters for the binomial tree of the original case (investment in 1998) Parameter Value (1) Stock price S0 3 634 194.92 € (2) Discount rate r 10% (3) Risk-free rate rf 5% (4) Volatility σ 30% (5) Exercise price 1 X1 808 838.00 € (5) Exercise price 2 X2 808 838.00 € (5) Exercise price 3 X3 808 838.00 € (5) Exercise price 4 X4 996 853.50 € (5) Exercise price 5 X5 996 853.50 € (6) Duration stage-1 T1 1 (6) Duration stage-2 T2 1 (6) Duration stage-3 T3 1 (6) Duration stage-4 T4 1 (6) Duration stage-5 T5 1 (7) Incremental time step δt 1 (8) u = exp (σ√δt) 1.34986 (8) d = 1/u 0.7408 (9) p = (exp (rfδt) – d)/(u-d) 0.5097 (1) S0 is the value of the underlying asset, calculated by discounting to 1998 all of the NPV of every single CSP plant using Eurotrough built worldwide from table 2. (2) Discount rate (r) is the opportunity cost of the company undertaking the investment. There is a sensitivity analysis based on discount rate for being one of the parameters that mostly affects the results. GMB Doctoral Thesis 71/130 (3) Risk-free rate (rf) is the discount rate assumed to be the riskless discount rate on the basis of the U.S. Treasury spot rate of return, with maturity equivalent to the option’s time to maturity. (4) Volatility (σ) is the risk of the expected returns. The market risk has been eliminated using a technology that has been already implemented. There is a risk associated with the performance of the CSP plant, however, depending on the weather, the availability factor of the plant, variations in the price of labour or certain commodities, such as water for cleaning mirrors. In this case, a volatility of 30% is determined. Nonetheless, there is also a sensitivity analysis of the volatility. (5) Exercise price of period i (Xi) means the cost of undertaking stage i. The R&D project has been divided into 5 stages (each being one year) in which the company can decide whether to continue (exercising the option at price Xi) or abandon at the current stage. The cost of each period is extracted from table 1, assuming Eurotrough I is undertaken in 1998, 1999 and 2000 (with the same investment each year, although the period in 1998 is shorter), and Eurotrough II is undertaken in 2001 and 2002. (6) Duration stage (Ti) is the length of each stage. It is assumed to be one year, indicating that the company could abandon the project at the beginning of each year following the stage gate approach (Cooper et al. 2018). (7) Incremental step (δt) is the discrete periods into which the model is divided. In this case, the step is one year. (8) u and d are the RO parameters applied to the binomial tree according to (Kodukula et al. 2006) y (Hull 2009). S0 is the initial value of the asset, but in the first time increment, it could either go up or down and from there continues to go either up or down in the following time steps. The up and down movements are represented by u and d factors, where u>1 and d=1/u. The magnitude of these factors depends on the volatility of the underlying asset. GMB Doctoral Thesis 72/130 (9) p is the risk-neutral probability, which makes it possible to use the risk-free rate as a discount rate according to (Kodukula et al. 2006; Hull 2009). 5.3 Scenario analysis In this section, the objective is to compare the original investment (1998 with 50% grant) with different volatility and risk scenarios in 1996 if the EC had tried to boost the investment. 5.3.1 Original investment in 1998 The original investment was decided in 1998. At that time, with the costs of table 1, the revenues from table 2 and the discount rate from table 3, the NPV of the project was nearly zero. Due to the positive effect of the grant, however, the NPV rose to 1.79 M Eur., meaning a ROI of approximately 40%. The detailed cash flow is included in Table 4. In (Martín-Barrera et al. 2016), there was a study of the positive financial effects of the grant versus the obligations imposed by signing a contract with the EC to develop the project. In fact, a minimum grant of 17.25% was determined as the break-even point between the improvements due to the EC contributions minus the value lost because of the lack of flexibility. 5.3.2. What would be the required grant to bring the project forward to 1996? According to Brown et al. (Brown et al. 2012), because R&D knowledge spills across firms and even countries, suggesting that socially optimal rates of R&D are likely to be much higher than privately optimal levels, R&D faces financing constraints such that its private investment may be below the optimal level. In GMB Doctoral Thesis 73/130 Europe, there are funds to finance sustainable energy projects to overcome market failures (Streimikiene 2005). For any funding body, and for society in general, it is advantageous to research and develop new technologies as soon as possible to boost the technological potential of the country and help its economic growth. The further from the market a new idea is, however, the riskier for the promoter. From this point forward, we will assume that even the EC wants the consortium to develop Eurotrough a few years earlier; the first commercial plant is still built in 2006. This assumption is the most negative for the company because revenues are received much later than the investment. However, we have decided to explore only the EC initiative to bring the inversion forward to 1996, without any side benefits, such as receiving revenues sooner. By doing so, the task is only focused on the grant. It is reasonable to think that the sooner the R&D project is finished; the sooner a new commercial power plant will be built. Therefore, if the R&D project had been performed in 1996, the first commercial power plant could have been built before 2006. This idea will be explored in greater detail at a later point. The selection of 2 years is considered reasonable because a 1-year period has a moderate effect and 3 or more years have a deep impact on the parameter estimation because a much higher volatility applies. It is unlikely that companies could start much earlier than 2 years simply due to the state of the art. With these values, the NPV in 1996 with everything else equal (see cash-flow in table 5) would be 1.16 M Eur. This assumption is not realistic, however, as there is a higher risk involved because the new trough is being developed 2 years earlier. To better model the investment in 1996, the discount risk should be higher. A reasonable figure is 13%, although sensitivity analyses are performed in this GMB Doctoral Thesis 80/130 Table 12: Sensitivity analysis of the Discount Rate Discount Rate gr’ ROV (σ=40%) gr’’ Difference 10% 68% 890.109 € 43% 25% 11% 77% 693.771 € 57% 20% 12% 86% 528.286 € 70% 16% 13% 94% 386.768 € 82% 12% 14% 101% 275.990 € 92% 9% 15% 107% 184.770 € 102% 5% In conclusion, the required grant increases almost 12% per each 1% increase in the discount rate. 5.5.2 Sensitivity analysis: volatility The volatility impact, however, is lower than the discount rate. Using a similar notation as before: gσ’: the grant required in 1996 to equal the NPV in 1998 with a 50% grant, assuming the discount rate is 10% in 1998 but in 13% in 1996. Because volatility only affects ROV and this value already includes flexibility in the analysis, the value is the same regardless of the change in σ. ROV (r=13%): Real Option Value (i.e., the value of the flexibility) depending on the volatility using a discount rate of 13%. The higher the volatility, the greater the flexibility value and therefore the ROV. gσ’’: the grant required in 1996 to equal the NPV in 1998 with a 50% grant, including the value of flexibility. As stated before, the effect is lower compared to the discount rate. GMB Doctoral Thesis 81/130 In conclusion, the required grant decreases 6% per 10% increase in volatility, as this table shows. Table 13: Sensitivity analysis of the Volatility σ gσ’ ROV(r=13%) gσ’’ Difference 30% 94% 180.446 € 88% 6% 40% 94% 386.768 € 82% 12% 50% 94% 585.338 € 76% 18% 5.3.2 What if the first commercial plant were built sooner? Although up to this point it has been assumed that the first commercial power plant was still built in 2006, if the R&D project began in 1996 rather than 1998, it is not unreasonable to assume that the first commercial power plant including this new component would have been built in 2004 and not in 2006. Assuming a discount rate of 13% and volatility equals 40% (same values as the base case), the NPV in 1996 would have been 0.90 M€ and the ROV = 0.387 M€. This flexibility value is close to the original case because there are two factors in this value: - A positive effect of S0 because it is discounted from 2004 rather than 2006; therefore, the initial value is greater. - A negative effect because the time to exercise the options is decreased by two years. In this particular case (r=13%, σ=40%), there seems to be a balance between the two factors. Obviously, if the model were calculated with r=10% and σ=40%, the ROV would be exactly the same as the original case in 1998 because the flexibility would have GMB Doctoral Thesis 82/130 been the same. These results show there is little correlation between the required grant and the commercial power plant being built earlier. Therefore, it is safe to continue using the 2006-year with no major impact on the calculations. 5.6 Combining parameters The previous sub-sections have performed a sensitivity analysis based on two individual parameters, the discount rate and the volatility. For the sake of providing a complete overview, however, this section compares three different scenarios. The notation is the same as before: g’ is the required grant in each scenario to equal the NPV in 1998 with 50% grant. After valuing the flexibility, the new required grant g’’ is lower. The results are shown in the below table. Table 14: Comparison of three different scenarios Scenario Discoun t Rate Volatility 1st Commercial Revenue g’ ROV g’’ Base r = 13% σ = 40% 2006 94% 386.768 € 82% Optimistic r = 10% σ = 50% 2004 50% 1.100.156 € 19% Pessimistic r = 15% σ = 30% 2006 107% 83.457 € 105% The main takeaway of this section is that the great value associated with flexibility depends on the discount rate; to a lesser degree, on volatility; and finally, on the commercial year. This finding creates no new problems with valuing R&D projects, however, as the discount rate is also the main factor in any conventional DCF (Brealey et al. 2008). GMB Doctoral Thesis 83/130 5.7 Discussion The case presented is a real case undertaken by a consortium of private companies that decided to start an R&D project in 1998 with a certain amount of grant funding (50% of the total cost). This facilitated a minimum NPV in the investment, taking into account the uncertainties associated with any R&D project, namely technical success and market demand. According to the market estimations and the expected NPV presented in (Martín- Barrera et al. 2016), the project would be undertaken only with a grant because the NPV without any grant was close to zero. This grant helps companies to overcome market inefficiencies regarding R&D (Streimikiene et al. 2007; Pennings et al. 1997). If the financing body, for instance the EC, wanted to incentivise the companies to start the project earlier, however, thus bringing forward the commercial plant and eventually increasing competitiveness, a larger grant would be necessary to balance the higher risks borne by the companies. To equal the NPV of 1998 in 1996, the grant would need to be not 50% but 94%. 1 This creates two drawbacks: a) Clearly, it is a very expensive measure and it would not be possible to bring the project forward more than a couple of years. Almost duplicating the grant would mean that half of the companies would be able to receive incentives. b) The closer a grant is to 100%, the higher the risk of projects with less commercial prospects going forward. For instance, a large R&D company or an R&D centre with no real commercial interest could apply for the grant for the sake of keeping their employees busy while covering structural costs. If there is no real risk for the company and consequently no commitment on the company’s 1 94% in the base case, Subsection 4.3.4.1 performed a sensitivity analysis and this grant could be a slightly lower or even greater. The rationale of the discussion remains the same. GMB Doctoral Thesis 84/130 part, it is difficult to estimate whether the project has any feasible expected profitability. The previous drawbacks are based on a non-flexibility scenario. As discussed earlier, however, the company has the option to defer the development as well as to abandon the project if the estimations are not promising. Including the value of the flexibility given by the results in table 7, the grant can be lowered. In fact, from the required 94% to equal the NPV in 1996 at a higher discount rate, an equivalent grant of 82% can be offered. The difference between 94% and 82% is precisely the value of the flexibility. As previously stated, this value is based on a very conservative scenario and is subject to reduction as the flexibility increases. Some findings regarding this statement are as follows: i) If the R&D project starts earlier, it is reasonable to assume the first commercial power plant will also be built earlier. This effect would increase the value of the underlying asset, as subsection 4.3.4.3 showed (albeit with a moderate effect). ii) If the above reasoning were not true, this would mean the project has a higher volatility value than the one used in the numerical case. A higher volatility means a greater ROV and a lower required grant in 1996. Subsection 4.3.4.2 determined a reduction of 6% in the required grant for every 10% increase in volatility. In fact, this relationship would be greater if the project’s start date was advanced by more than 2 years because the impact of the volatility is increased with the length of the options. iii) If neither i) nor ii) were true, it would mean the risk is not so high, and therefore, the discount rate would also decrease. In this case, similar to i), the underlying asset value would not only increase but would be boosted. Subsection 4.3.4.1 determined a reduction of 12% in the required grant per 1% decrease in the discount rate. GMB Doctoral Thesis 85/130 Chapter VI. Conclusions Flexibility allows some of the uncertainty to be resolved before irreversible expenditures are made, enabling managers to make decisions that bring their operations closer to maximizing profits GMB Doctoral Thesis 86/130 The current paradigm in the valuation and assessment of R&D projects has been questioned because their level of flexibility and embedded options. Several studies have demonstrated how ROA should be applied in this kind of projects, developing appropriated frameworks to evaluate projects of different R&D areas. The uncertainty of R&D projects, the spillover benefits R&D expenditures provide to society and the increased risk and reduced cost of capital have justified governments implement public policies to subsidy, and ROA has been used for evaluation of subsidy schemes of several countries. But obtaining public funding imposes loss of flexibility in the management of its projects for companies once a grant is awarded. These restrictions have to be considered in assessment of projects and policies. Moreover, one of the most debated question in the literature has been if public incentives stimulate (crowding-in) or reduce private investment (crowding-out). This work presents the implications in assessment of these restrictions for companies and governments in two empirical studies for the same project but with different perspectives, using ROA. The first case (chapter 4) proved not any subsidy should be enough for private companies. On the one hand, grants improve cash-flow via incomes during the R&D investment period. On the other, they reduce the total value of the R&D project because of the stated limitations. The value added to a project by a grant is significantly lower than its nominal amount due to the loss of flexibility for the manager. In this case, stage-gate is useless because flexibility, which has proved to represent a significant share of the project (Santos et al. 2014; Martínez-Ceseña et al. 2011; Davis et al. 2003), is non-existent. Therefore, a minimum acceptable grant (a break-even point) that equals the loss in flexibility has been calculated. If some options are included in the incentive program, there will be less lost value due to inflexibility and a greater minimum grant acceptable. So, flexible grant schemes could provide higher value for companies, although the economic subsidies would be lower. The analysis framework used in ROA has to include the options restrictions because of the government grant; this provides a more accurate GMB Doctoral Thesis 87/130 project value. On the contrary to NPV, in which any additional income will always improve the valuation of the project, the reality is that grants below a threshold imply a loss of value for the investment, as they impose certain constrains limiting the company managerial flexibility. Companies need to balance on one hand the additional income due to the grant and on the other the limitations imposed (such as setting a deadline, not allowing to abandon the project so the investment is irreversible, discarding switching to alternative technologies, etc.) and realize the underlying value on flexibility could be much greater that the incentive. When a policy R&D incentive is designed, it should be recognized that financial terms are considered by companies facing an R&D investment as well as side effects such as the flexibility permitted during development. Some works analyse policy evaluation models from the perspective of governments and investors (Zhang et al. 2014). However, this approach will be useful to policy makers defining new public R&D financing policies because it provides a new way to quantify a concern faced by many companies. A new variable has been added to the debate regarding whether public R&D financing crowds-in or crowds-out private spending: the flexibility of the grant scheme. This new idea adds to the current literature and offers tools for policy makers that will help shape a more complete public R&D policy. For these reasons, the second case (chapter 5) set the focus on the rationale for public institutions. They should incentive the promotion of R&D projects but without displacing (ie. crowding-out) the private investment. Then, rather than providing very large grants which also presents other drawbacks such as freeriders; public institutions could think about alternative schemes with a lower grant but with much more flexibility. Thus, this new grant scheme could be equally attractive for private companies although the nominal grant is substantially lower. Grants can be divided in two terms: the well-known financial incentive, the only one used to date; and a new term, the freedom given to companies. This freedom, GMB Doctoral Thesis 88/130 namely flexibility, enhances the R&D project because companies do have different options over the course of the project such as abandoning the project at any time if the technical or economic feasibility is not clear; deferring development until the market demand or some technical barriers are clarified; switching to a more promising technology, and so on. This flexibility has been quantified using Real Options Valuation in order to determine the reduction in the grant size. This work has proved it is equally attractive for companies to have a traditional high (in monetary terms) grant that is very restrictive with almost no flexibility at all or the new paradigm of a lower grant with embedded options, for instance, the opportunity to abandon the project at any time or defer the project several years. This finding helps to contribute the existing debate regarding the size of the grant and its crowing-out effect. It seems reasonable to argued that a lower grant that shares the risk with private companies brings some advantages: (i) Because the grant is smaller, the funding body can reach more projects and companies with the same budget, facilitating long-term stability in public R&D funding policy that is less dependent on yearly national budgets. (ii) Companies could be more sincere and honest with the outcomes of R&D projects. Currently, abandoning a R&D project has two immediate consequences: returning the incentive received and reducing the probability of receiving public money in the future due to the reputation loss. Because there is an information asymmetry between the private company and the funding body, current grant schemes encourage the temptation not to abandon certain R&D failures in order not to harm the firm’s ability to be awarded grants in the future. (iii) As stated above, the lower the grant is, the less crowding-out effect it represents. According to the literature, the crowding-out effect appears to GMB Doctoral Thesis 89/130 be correlated with large grants. Therefore, lower grants are less likely to negatively displace private R&D investment. (iv) Finally, the risk-sharing profile obliges companies to be selective with truly feasible R&D projects. Schemes in which financing reaches 100% of the investment may incentivise free-riders whose sole goal is to receive grants to maintain R&D structures without any real commercial potential. This second study has presented two scenarios a funding body may face at any time: on the one hand, reducing the nominal grant but awarding it in the same year, and on the other, deciphering the impact of awarding the grant earlier when the company faces higher risk in alignment with the goal of bringing knowledge to society sooner without bearing all the cost of the R&D project. After the calculations, a sensibility analysis over the design parameters of R&D activity was performed proving that Real Options Valuation, as DCF, is mostly affected by the discount rate, so no new significant variables are introduced to the valuation. Other parameters such as volatility have a much lower impact on the equivalent grant calculation. In summary, this doctoral thesis has proven the following statements: • NPV techniques are not valid for R&D project valuation under grant schemes. • Companies should require a grant above a minimum threshold to balance the loss of flexibility imposed by the grant requirements. • Public institutions should never provide 100% grants; but lower and more flexible incentives which are equally attractive for companies. • Then, there is an optimal balance between 0% and 100% that will depend on how mature is the project and therefore the managerial flexibility still embedded. GMB Doctoral Thesis 96/130 Kumbaroğlu G, Madlener R, Demirel M. A real options evaluation model for the diffusion prospects of new renewable power generation technologies. Energy Econ 2008;30:1882-1908. Lach S. Do R&D Subsidies Stimulate or Displace Private R&D? Evidence from Israel. J Ind Econ 2002;50:369–90. doi:10.1111/1467-6451.00182. Lee S, Shih L. Enhancing renewable and sustainable energy development based on an options-based policy evaluation framework: Case study of wind energy technology in Taiwan. Renewable and Sustainable Energy Reviews 2011;15:2185- 2198. Lee S, Shih L. Renewable energy policy evaluation using real option model — The case of Taiwan. Energy Econ 2010;32, Supplement 1:S67-S78. Lin B, Wesseh Jr PK. Valuing Chinese feed-in tariffs program for solar power generation: A real options analysis. Renewable and Sustainable Energy Reviews 2013;28:474-482. Lööf H, Heshmati A. The Impact of Public Funding on Private R&D investment: New Evidence from a Firm Level Innovation Study. Work Pap Ser Econ Institutions Innov 2005;6:1–26. Marino M, Parrotta P, Sala D. New Perspectives on the Evaluation of Public R&D Funding. University of Aarhus, Aarhus School of Business. Available from: Department of Economics [online]. Martín-Barrera G, Zamora-Ramírez C, González-González JM. Application of real options valuation for analysing the impact of public R&D financing on renewable energy projects: A company′s perspective. Renew Sustain Energy Rev 2016;63:292–301. doi:10.1016/j.rser.2016.05.073. GMB Doctoral Thesis 97/130 Martínez Ceseña EA, Mutale J, Rivas-Dávalos F. Real options theory applied to electricity generation projects: A review. Renewable and Sustainable Energy Reviews 2013;19:573-581. Martínez-Ceseña EA, Mutale J. Application of an advanced real options approach for renewable energy generation projects planning. Renewable and Sustainable Energy Reviews 2011;15:2087-2094. Menegaki A. Valuation for renewable energy: A comparative review. Renewable and Sustainable Energy Reviews 2008;12:2422-2437. Merton RC. Theory of rational option pricing. The Bell Journal of economics and management science 1973:141-183. Meuleman M, De Maeseneire W. Do R&D subsidies affect SMEs’ access to external financing? Res Policy 2012;41:580–91. doi:10.1016/j.respol.2012.01.001. Mohnen P, Palm FC, van der Loeff SS, Tiwari A. Financial Constraints and Other Obstacles: are they a Threat to Innovation Activity? Economist (Leiden) 2008;156:201–14. doi:10.1007/s10645-008-9089-y. Monjas-Barroso M, Balibrea-Iniesta J. Valuation of projects for power generation with renewable energy: A comparative study based on real regulatory options. Energy Policy 2013;55:335-352. Myers SC. Determinants of corporate borrowing. J Financ Econ 1977;5:147-175. National Renewable Energy Laboratory (NREL). Concentrating Solar Power Projects - Parabolic Trough Projects 2014 [online]. Available from: 〈http://www.nrel.gov/csp/solarpaces/parabolic_trough.cfm〉; 2014. Ollila J. Real Options in Pharmaceutical R&D. Master's Thesis, Helsinki University of Technology 2000. Özçelik E, Taymaz E. R&D support programs in developing countries: The Turkish experience. Res Policy 2008;37:258–75. doi:10.1016/j.respol.2007.11.001. GMB Doctoral Thesis 98/130 Pennings E, Lint O. The option value of advanced R&D. Eur J Oper Res 1997;103:83–94. doi:10.1016/S0377-2217(96)00283-4. Reuter WH, Szolgayová J, Fuss S, Obersteiner M. Renewable energy investment: Policy and market impacts. Appl Energy 2012;97:249-254. Ross SM. An Elementary Introduction to Mathematical Finance: Options and Other Topics Cambridge University Press; 2003. Santos L, Soares I, Mendes C, Ferreira P. Real Options versus Traditional Methods to assess Renewable Energy Projects. Renewable Energy 2014;68:588-594. Sarkis J, Tamarkin M. Real Options Analysis for Renewable Energy Technologies in a GHG Emissions Trading Environment. In: Anonymous Emissions Trading, Springer; 2008, p. 103-119. Setter O, Tishler A. Investment policies in advanced defense R&D programs. Tel Aviv: 2005. Siddiqui A, Fleten S. How to proceed with competing alternative energy technologies: A real options analysis. Energy Econ 2010;32:817-830. Siddiqui AS, Marnay C, Wiser RH. Real options valuation of US federal renewable energy research, development, demonstration, and deployment. Energy Policy 2007;35:265-279. Stiglitz J, Weiss A. Credit rationing in markets with imperfect information. Am Econ Rev 1981;71:393–403. Streicher G, Schibany A, Gretzmacher N. Input Additionality Effects of R&D Subsidies in Austria: Empirical Evidence from Firm-level Panel Data. Wien: 2004. Streimikiene D, Klevas V, Bubeliene J. Use of EU structural funds for sustainable energy development in new EU member states. Renew Sustain Energy Rev 2007;11:1167–87. doi:10.1016/j.rser.2005.07.006. GMB Doctoral Thesis 99/130 System Advisor Model (SAM) 2014 [online]. Available from: https://sam.nrel.gov/〉; 2014. Trang NT, Takezawa N, Takezawa N. Real options and the evaluation of research and development projects in the pharmaecutical industry: a case study (Special Issue on Theory, Methodology and Applications in Financial Engneering). Journal of the Operations Research Society of Japan 2002;45:385-403. Trigeorgis L. Real options: Managerial flexibility and strategy in resource allocation. MIT Press; 1996. Tsui M. Valuing innovative technology R&D as a real option: application to fuel cell vehicles. Doctoral dissertation. Massachusetts Institute of Technology; 2005. Ughetto E. Does internal finance matter for R&D? New evidence from a panel of Italian firms. Cambridge J Econ 2008;32:907–25. doi:10.1093/cje/ben015. Venetsanos K, Angelopoulou P, Tsoutsos T. Renewable energy sources project appraisal under uncertainty: the case of wind energy exploitation within a changing energy market environment. Energy Policy 2002;30:293-307. Wang T, De Neufville R. Real options “in” projects 2005. Wiser RH, Pickle SJ. Financing investments in renewable energy : the impacts of policy design. Renew Sustain Energy Rev 1998;2:361–86. doi:10.1016/S1364- 0321(98)00007-0. Wolff GB, Reinthaler V. The effectiveness of subsidies revisited: Accounting for wage and employment effects in business R&D. Res Policy 2008;37:1403–12. doi:10.1016/j.respol.2008.04.023. Yu W, Sheblé GB, Lopes JAP, Matos MA. Valuation of switchable tariff for wind energy. Electr Power Syst Res 2006;76:382-388. GMB Doctoral Thesis 100/130 Zhang M, Zhou D, Zhou P. A real option model for renewable energy policy evaluation with application to solar PV power generation in China. Renewable and Sustainable Energy Reviews 2014;40:944-955. Zhang X, Wang X, Wang X, Chen H. Energy uncertainty risk management of hydropower generators 2005:1-6. Zúñiga-Vicente JÁ, Alonso-Borrego C, Forcadell FJ, Galán JI. Assessing the effect of public subsidies on firm R&D investmen: A survey. J Econ Surv 2014;28:36–67. doi:10.1111/j.1467-6419.2012.00738.x. GMB Doctoral Thesis 101/130 Appendix I Application of real options valuation for analysing the impact of public R&D financing on renewable energy projects: A company ′ s perspective GMB Doctoral Thesis 102/130 GMB Doctoral Thesis 103/130 GMB Doctoral Thesis 104/130 GMB Doctoral Thesis 105/130 GMB Doctoral Thesis 112/130 GMB Doctoral Thesis 113/130 GMB Doctoral Thesis 114/130 GMB Doctoral Thesis 115/130 GMB Doctoral Thesis 116/130 GMB Doctoral Thesis 117/130 GMB Doctoral Thesis 118/130 GMB Doctoral Thesis 119/130 GMB Doctoral Thesis 120/130 GMB Doctoral Thesis 121/130 GMB Doctoral Thesis 128/130 Figure 12: RSER is ranked #11 under Engineering and Computer Science by Google Scholar according to h5 index CiteScore and percentile Source: https://www.scopus.com/sourceid/27567?origin=sourceInfo&zone=refpointrank CiteScore 2015 counts the citations received in 2015 to documents published in 2012, 2013 or 2014, and divides this by the number of documents published in 2012, 2013 and 2014. Figure 13: RSER is ranked in the 96th percentile of Renewable Energy, Sustainability and the Environment according to CiteScore, being ranked #5 out of 117 journals. Citations received GMB Doctoral Thesis 129/130 The first paper was published in September 2016 and has already received the following two citations: Figure 14: Citations received on Part 1 of this doctoral thesis Figure 15: First citation received, published in RSER in November 2016 GMB Doctoral Thesis 130/130 Figure 16: Second citation received, to be published in Renewable Energy in August 2017