A critical analysis of recent advances in the techniques for the evaluation of renewable energy projects

Available online at www.sciencedirect.com

ScienceDirect International Journal of Project Management 31 (2013) 1057 – 1067 www.elsevier.com/locate/ijproman

A critical analysis of recent advances in the techniques for the evaluation of renewable energy projects Chen-Yu Chang ⁎ Bartlett School of Construction and Project Management, University College London, 1–19 Torrington Place, London WC1E 7HB, United Kingdom Received 21 June 2012; received in revised form 28 February 2013; accepted 5 March 2013

Abstract Renewable energy investment is an integral part of sustainable economic development agenda. Whereas some important advances have been made in recent years to assist project investors in making better use of financial risk management instruments and taking into consideration real options embedded in renewable energy projects, this research asserts that, owing to failure to consider both behavioural uncertainty and the limit of risk transfer, these approaches may still lead to a biased evaluation result. Drawing on a novel concept of “risk-bearing capacity”, the research suggests developing a new approach whereby investors can incorporate the choice of financial protection measures into investment evaluation in a coherent way. © 2013 Published by Elsevier Ltd. APM and IPMA. Keywords: Risk management; Sustainability; Project evaluation; Transaction cost; Behavioural uncertainty

1. Introduction On the back of growing concern over climate change and high volatility of oil prices, investment in renewable energy 1 (RE) has been rising at a compound rate of 30% since 2004 and reached the scale of £243 bn in 2010 globally. The development of renewable energy is now deemed as an integral part of sustainable economic development in the future (WEF, 2011). In spite of its importance, investment community still cannot fully grapple with the underlying risk involved in this new investment class (EDHEC, 2010). This is why in recent years we have seen many approaches being developed to address this gap. The purpose of this paper is to give a review of these advances by critically examining their limitations from the perspective of risk-bearing capacity.

⁎ Tel.: + 44 20 76791266. E-mail address: [email protected]. 1 Generally, renewable energy (also known as new energy) sources refer to solar photovoltaic, solar thermal, biofuels, wind energy, wave energy, tidal, hydro, and geothermal, but this research only focuses on two of the most important types: solar photovoltaic and wind. 0263-7863/$36.00 © 2013 Published by Elsevier Ltd. APM and IPMA. http://dx.doi.org/10.1016/j.ijproman.2013.03.001

Generally, a RE project has three characteristics: First, the cost disadvantage of renewable energy makes it highly dependable on government subsidies to compete with traditional energy. The viability of a RE project therefore strongly depends on the price at which its output is sold to the national grid (known as feed-in tariff). This regulatory risk is not in the investor's control. Second, one of the greatest appeals of renewable energy is its ability to tap into free natural resources as fuel, such as solar and wind power. However, this advantage is offset by greater exposure to the uncertainty of power supply. For instance, how many hours a day wind is strong enough to propel turbines would vary across the project life. Third, the technological knowhow needed for the production of renewable energy is often controlled by a small group of vendors. RE investors usually find it desirable to bundle the functions of Engineering design, Procurement of machinery and Construction into a single contract (known as EPC contract). These characteristics jointly shape the risk profile of RE projects that needs to be carefully evaluated in making investment decisions. To improve the quality of RE investment, we have seen two significant advances in recent years. United Nations Energy

1058

C.-Y. Chang / International Journal of Project Management 31 (2013) 1057–1067

Programme (UNEP) commissioned a series of reports for exploring the way financial risk management (FRM) instruments (e.g., insurance and derivatives) can be applied to reduce RE investors' cost of capital and thus facilitate investment in this field. Researchers have also made a stride in the development of new appraisal techniques. Unlike the traditional net present value approach, the value of options to change course of actions at later stage is increasingly acknowledged in project appraisal for two reasons: first, complicated mathematics involved in the pricing of real options has been a barrier to it being a practical appraisal tool, but the barrier was gradually lifted owing to the acceptance of its numerical solutions; second, real options analysis is flexible enough to work alongside other decision-aided methods, such as decision tree analysis and system dynamics, which enables researchers to broaden its applications. Despite these progresses, some crucial issues are still being left out of analysis. First, RE projects normally involve the employment of an integrated procurement system (EPC contract being a case). For a project investor (she, henceforth), the benefit of delegating the responsibilities to a single contractor (he, henceforth) is to reduce the effort that she must otherwise make for coordinating agents and to increase the transparency of liability for quality defects. However, single point of responsibility would aggravate the hold-up problem (Ive and Chang, 2007). In economics and management studies, a rich body of theoretical and empirical literatures evince that one party's intention to hold up the other would turn a trading relationship into a confrontational one, making both of them incur additional but unnecessary costs in resolving disputes. These costs, termed transaction costs, will impinge upon the efficiency of transactions and inhibit economic activity (Hart, 1995; Williamson, 1985, 1996). In stark contrast to its prominence in the literature, this problem is not considered in the mainstream project risk management guides as yet (ICE, 1998; Simon et al., 1997). Whilst we have seen some awareness of the importance of behavioural uncertainty in the study of risk management (Dailami et al., 1999) and supply contract design (Ritchie and Brindley, 2007), there is still a long way to reach fruition. Second, the primary purpose of project appraisal is to inform project investors of the value an investment opportunity would create. Generally, the existing methods, including the Net Present Value (NPV) approach and real options analysis, centre on the evaluation of intrinsic value of a project as risks are only considered in the “raw” form. Theoretically and practically, it is well recognised that FRM instruments can help reduce variability in project cash flows with the effect of lowering the project's cost of capital and raising its economic and financial viability. UNEP's attempt at facilitating RE investments through a better choice of risk management products is a pertinent case (UNEP, 2004, 2006, 2007, 2008). As a result, an appraisal method accounting for the net effect of risk handling measures on project cash flows (i.e., risks treated in their “processed” form) should be able to produce an evaluation result of greater reference value. Third, in recent years, a host of high-profile project failures in the UK has resulted in a waste of massive taxpayers' money (NAO, 2001, 2004, 2006, 2009). A central lesson we can learn from these projects is that a contract can only withstand a limited

amount of risk, so it is necessary to take contract breakup as a real danger in construction contracting. For these reasons, this research will, on the one hand, conduct a critical review of the latest development of evaluation techniques with focus on their mechanics and, on the other, explore feasible ways to address the deficiencies identified. As will be shown later, this research suggests that a novel concept, termed risk-bearing capacity, can offer a coherent way to incorporate the effect of contracting hazards and the choice of FRM instruments into an integrated framework of project evaluation. This new line of inquiry would hold promise to open a new frontier for the studies of project appraisal and project risk management in the future. 2. A literature review 2.1. An overview of the existing approaches Evaluation of RE projects has attracted extensive research interest in recent years. The approaches employed can be grouped into three categories: 2.1.1. The NPV approach This approach is a standard technique in the appraisal of RE projects in practise (Owens, 2002) and in research alike. Ozerdem et al. (2006) analyse three cases where investment is made by an auto producer, auto producer group, and independent power producer respectively, and compare their cost advantage and profitability in terms of the present cost per kWh and the internal rate of return. Falconett and Nagasaka (2010) apply Monte Carlo simulations to model the impact of risks using the NPV approach to explore the effect of alternative support mechanisms (feed-in tariffs or renewable energy certificate) on the financial feasibility of small-scale RE projects. For investments involving a portfolio of RE projects, Torriti (2012) develops a method to allow for multiple-project discount rates to be used to reflect the intertemporal profile of project risks. In these works, the pivotal role of FRM instruments in enhancing the viability of a RE project is not addressed. By contrast, a series of studies conducted by the UNEP take it as a central issue in an effort to build a NPV based simulation model to incorporate the choice of financial protection into the appraisal decision (UNEP, 2007). This method is chosen for further review in Section 2.2 because of its ability to consider comprehensive factors and its greater readiness to be accommodated in a practical decision-making scenario. 2.1.2. Multi-criteria decision-making approach This approach has been called on to analyse the choice of energy technologies and energy projects (Huang et al., 1995; Siskos and Hubert, 1983; Zhou et al., 2006). By attaching weightings to reflect their importance based on expert evaluation, this method enables decision makers to take all relevant factors into account. In the evaluation of the relative desirability of four types of RE sources (wind, hydroelectric, solar and biomass) as the means to achieve Spain's national target of 12% of primary energy being generated by RE sources, San Cristóbal (2011) employs a multi-criteria ranking method (known as the VIKOR

C.-Y. Chang / International Journal of Project Management 31 (2013) 1057–1067

method) to account for the effect of wide-ranging factors, including power, investment ratio, implementation period, economic life, operating and maintenance cost, as well as carbon emission reduction. Whereas this method is useful in the planning of RE investment (Borges and Antunes, 2003; Haralambopoulos and Polatidis, 2003), it would run into problems when the factors involve contracting issues. For example, in the application of the multiple-attribute approach to the selection of procurement systems, a major deficiency lies in the capability of design–build and fixed-cost contracts in offloading risk (Chang and Ive, 2002). Most of the literatures implicitly assume that contracts are complete, whereby the ex ante commitment can always be delivered as promised. However, evidence shows it is not the case in reality (Contracting Excellence, 2004). What is more, this approach works in a methodological framework distinct from both the NPV approach and real options analysis, so it is not reviewed in further details. 2.1.3. Real options analysis Competition in the wake of deregulation of electricity markets makes it appreciably important for energy investors to factor in “flexibility” in the implementation of energy projects (Felder, 1996; Venetsanos et al., 2002). Under this trend, the appraisal bias resulting from ignoring the value of options to change (e.g., defer, expand, and switch) would grow large. It is the main reason why an increasing application of real options analysis has been made to the analysis of different aspects of RE investment, such as diffusion prospects of RE technologies (Kumbaroğlu et al., 2008), RE policy evaluation (Boomsma et al., 2012; Lee and Shih, 2010, 2011), and investment evaluation (Fernandes et al., 2011; Lee, 2011). A common feature of these studies is that they all follow the framework of traditional real options analysis. However, we have seen some innovative attempts made to accommodate decision analysis (Brandao and Dyer, 2005) and system dynamics (Tan et al., 2010) into the real options analysis. Since the overarching

1059

purpose of this paper is to find the way forward through a critical scrutiny of recent methodological advances, so, apart from a review of basic real options analysis in Section 2.3.1, this research will focus on these innovations in Section 2.3.2 and 2.3.3.

2.2. UNEP reports To increase RE investments in developing countries, UNEP has strived to explore the way FRM instruments can be utilised to reduce the capital cost of RE projects (UNEP, 2004, 2006, 2007, 2008). These reports suggest that use of FRM instruments can achieve three benefits: reducing the default rate to a minimum, producing an optimal credit rating, and achieving a high internal rate of return for the investor (UNEP, 2007). An overview of the method is shown in Fig. 1 (UNEP, 2007). The basic model is made up of basic project financial information (e.g., financial structure, tax assumptions, capital costs and purchase price of electricity) and perceived risk sources (e.g., annual energy output, construction delays and carbon credit sale price). The possible financial outcomes of the project are predicted using the Monte Carlo simulation. In each of 5000 simulations, shocks are randomly drawn from the underlying probability distributions and then fed into cash flow calculations. The output of simulations can yield a cumulative probability distribution of key performance metrics (e.g., equity IRR or project NPV). Project investors can arbitrarily choose a reasonable threshold as decision rules, such as “there is at least 95% that the project value should exceed a certain level”. The similar targets can be set against other metrics, such as equity IRR or default rate. The effects of FRM instruments are judged by the changes in the descriptive statistics of key performance metrics as a result of their use. For instance, project investors can track the mean or 90th percentile of equity IRR to assess whether financial protection in place is sufficient.

Fig. 1. UNEP framework for the analysis of the effect of FRM instruments in the evaluation of RE projects (UNEP, 2007).

1060

C.-Y. Chang / International Journal of Project Management 31 (2013) 1057–1067

Whereas this approach has wide applicability to RE projects, it is unsatisfactory in three aspects: first, the results based on different metrics might be inconsistent and no advice is provided on how to reconcile conflicting results; second, the selection of metrics, key statistics and thresholds is arbitrary; third, no allowance is made for contracting hazards and the limit of risk transfer. 2.3. Real options analysis 2.3.1. From the NPV approach to real options analysis It is widely accepted in academia that a capital project should be evaluated on the basis of the discounted value of its forecast cash flows. Inevitably, cash flow realisations may deviate from their central estimates, so the downside risk should be properly priced into project appraisal. In the NPV approach, the potential impacts of risk sources are “lumped” into the discount rate, which is meant to reflect the rate of return that the invested capital would otherwise yield in comparable investment opportunities elsewhere. The robustness of this approach therefore rests with the reliability of the rate chosen for discounting future uncertain cash flows. Theoretically, a reliable estimate of the discount rate can be obtained by using the weighted rate of return on the portfolio that can perfectly mimic the risk profile of the project (Brach, 2003). However, this solution has two limitations: first, the volatility of payoffs from the project may be also driven by private risks (Amram and Kulatilaka, 2000). For example, project investors may be lack of the ability to keep project cost under control or to select the right equipment to instal in the project. In principle, a project merely exposed to private risks should be discounted differently (Mun, 2002). However, the distinction between market risks and private risks is sometimes blurred, rendering the choice of discount rate difficult. Second, even for the projects merely exposed to market risks, the choice of discount rate is no easier. In circumstances where a replicating portfolio is not available, as is normally the case in practise, the project discount rate is benchmarked against the project investor's cost of capital. Where the new project is as risky as the investor's organisation as a whole, the investor's weighted cost of capital (WACC) is normally used as a proxy of the discount rate. However, if it is not the case, an adjustment can only be made by judgement. Apart from the difficulty of choosing an appropriate discount rate, the NPV approach has also been criticised as failing to consider the value of flexibility (Myers, 1984). In making investment decisions, investors may want to maintain the flexibility to change at later stages (e.g., deferral, expansion, and abandonment) when more information becomes available. Keeping options open for a period of time can sometimes create value. Drawing on the pricing model of financial options, researchers have endeavoured to develop methods for evaluating these “real” options. Unlike the NPV approach in which risks are reflected in a discount rate, real options analysis treats risks individually by computing their certainty equivalent for each period. The benefit of doing so is that decision makers can discount cash flows at a risk-free rate and evade the problem as to how to adjust the WACC to account for project risks.

There are three approaches to real options valuation: inspired by the seminal work of Black and Scholes (1973), the first approach values real options in the same way as financial option under the assumptions of replicating portfolio and no-arbitrage (Brennan and Schwartz, 1985). The idea is that the return of the project can be tracked by a portfolio of traded securities of comparable risk because any arbitrage opportunity will be picked up immediately in the market. However, the practicality of obtaining a market-priced portfolio to mimic project payoffs has posed a major barrier to the wide application of this approach. As a result, a variant of this approach suggested using a subjective estimate of project NPV and its volatility as inputs to the Black–Scholes equation (Luehrman, 1997; Luehrman, 1998a; Luehrman, 1998b). The second approach took a big step away from the financial options pricing model. Instead of searching for a “twin” security, it posits that we can take “the present value of the cash flows of the project without flexibility” as “an unbiased estimate of the market value of the project were it a traded asset” (Copeland and Antikarov, 2001). The reason seems straightforward. The asset that has highest correlation with the project in payoffs is the project itself (Howell et al., 2001). The benefit of making this so-called Marketed Asset Disclaimer (MAD) is that it considerably reduces the burden of data collection, making it no more demanding than the NPV approach, and broadens its application to virtually every investment decisions. The MAD approach also assumes that asset prices follow geometric Brownian motion, so that the option value can be solved using binomial lattices (Cox et al., 1979). Both of the two approaches discussed so far are built on the no-arbitrage assumption. However, no longer can this assumption hold when the project is exposed to private risks. The problem can be solved by incorporating two powerful methods in management sciences, dynamic programming (Dixit and Pindyck, 1994) and decision analysis (Amram and Kulatilaka, 2000), into the real options framework. Some significant efforts have been made on how to use decision analysis as a complimentary tool to valuing risk that cannot be hedged by trading (Brandao et al., 2005; Smith and McCardle, 1998; Smith and Nau, 1995). This line of inquiry prompted a growing interest among researchers in exploring a new way to relax the restrictive assumptions in real options analysis by drawing on the strength of decision analysis and system dynamics. 2.3.2. Real options embedded in decision tree analysis Cox et al. (1979) developed the first integrated procedure to evaluate real options by binomial decision trees. However, it is subject to a limitation that market risks sometimes cannot be distinguished from private risks (Brandao and Dyer, 2005). To make the approach more generally applicable, Brandao et al. (2005) and Brandao and Dyer (2005) proposed incorporating the MAD assumption into decision analysis and using binomial lattices to find the approximation solution to the underlying stochastic process. Suppose a project with unknown value V has a duplicating portfolio consisting of A units of traded stock with current price S and B dollars worth of a risk-free bond that offers a rate of return r. The value of the project will vary with the stock price

C.-Y. Chang / International Journal of Project Management 31 (2013) 1057–1067

that may go up with probability q to Su and go down with probability 1 − q to Sd. An interesting question is what probability (p) will a risk-neutral representative investor assign to the up state so as to make him feel indifferent to the return offered by the duplicating portfolio (AS + B) (see Fig. 2). The answer can be found by basic algebraic operations (Copeland and Antikarov, 2001): p¼

ð1−r−dÞ : ðu−dÞ

ð1Þ

With this probability, we can compute the risk-neutral measure of payoffs for each decision node based on forecast payoffs in each state of nature and discount these cash flows at a risk-free rate. The parameters needed for the calculation of p (i.e., u and d in Eq. (1)) are determined by the stochastic process underlying the value of the project. The Geometric Brownian Motion (GBM) employed in the Black–Scholes model is a common choice because it implies all the relevant information has been factored into the stock price at a point in time, which is normally the case (Hull, 2003). By making apffiffiffifficonvenient assumption that u = 1 / d, we can find u ¼ reσ Δt and thus p = (1 + rΔt − d) / (u − d). The project value at any point in time can be computed by Vi,j = V0u i − jd j (Copeland and Antikarov, 2001; Copeland and Tufano, 2004). Once the project value at each node is known, it is easy to calculate the NPV for each endpoint. The probability for each branch of a chance node should be determined according to the nature of the chance: for chance notes with private risks, use the investor's subjective probabilities; for chance nodes with market risks, use risk-neutral probabilities (Eq. (1)). The optimal option and its associated value can be worked out by “rolling back” the tree. 2.3.3. Real options analysis embedded in system dynamics Systems dynamics has merit in increasing the realism of the project model. Integrating system dynamics models of projects into real options analysis and decision analysis would create room for improving the quality of investment decisions. Tan et al. (2010) set analysis in the context of a RE project to demonstrate the way system dynamics algorithm can be fit into decision analysis whilst allowing for flexibility. Suppose a project investor considers building a 40 MW wind farm. There are two options in this project: (1) due to economy of scale, the capacity can be built to 50 MW at once or build 25 MW a year and expand into 50 MW later or just build 25 MW; (2) the start of construction can be postponed up to two years. The analysis starts with identifying the sequences of managerial decisions. Based on the project information and some engineering parameters, the standard system dynamics

q

p

Su

S

Vu

V 1-q

Sd

Fig. 2. Risk-neutral probability (p).

1-p

Vd

1061

model runs Monte Carlo simulations to evaluate the impact of three main risk sources identified in the project (natural gas prices, the learning effect of the wind farm supplier and the expiration date of the production tax credit) on cash flows. The probability distribution of cash flows at each chance nodes can be discretised to represent the likelihood of each option being chosen by applying the bracket median approximation technique (Clemen, 1997). This representation of chances has two advantages (Tan et al., 2010): first, it allows for joint impacts of multiple uncertainties whilst keeping the size of decision tree manageable; second, it offers a simplified solution for calculating path-dependent cash flows. The second advantage is particularly important because end-of-period NPV is dependent both on the decisions chosen and the chances eventuated on the path. What happen in the previous period will affect the cash flow in the current period. Therefore, the next step is computing conditional cash flow distribution for each decision sequence. With the aid of software DPL™ (decision programming language), conditional cash flow distributions and NPV at terminal nodes can be computed easily and the optimal decision path determined accordingly. 3. A risk-bearing capacity based approach: rationale and conceptual framework 3.1. Why quasi-rent is a good proxy for risk-bearing capacity As discussed so far, a major advance in the evaluation of RE projects lies in the distinction between project risks and market risks. Unlike market risks whose price can be reliably estimated with reference to traded securities of comparable risk, private risks arising from factors specific to individual firm and/or project should rely on subjective assessment. This is why it is beneficial to incorporate decision theory into the framework of real options analysis. However, the implication of the presence of private risks for project appraisal is more than what has been considered. Capital projects are mostly implemented through a contract. How well a contract can manage the whole contracting process is a decisive factor for how much value a project could create. The significance of this issue is evidenced by a vast volume of literature on post-contract contracting issues in economics. As is systematically elaborated in Bolton and Dewatripont (2005), economic approaches to these issues have two branches. Under the paradigm of complete contracting, a contract is modelled to be perfectly contingent-dependent whereby the financial consequence of contingencies can be well governed by the contract. Even in long-term contracting, trading parties can still write a renegotiation-proof contract to ensure the ex ante division of gains from trade be maintained ex post. However, this view is countered by the proponents of incomplete contracting paradigm. It is argued that, subject to contract drafter's limited foresights, the contract can only rule in, but not rule out what will happen under the contract (Hart and Moore, 2008). Contractual incompleteness makes it impossible to preclude the occurrence of ex post expropriation of the reward an investing party expects to reap at the end of the transaction. Currently, the literatures on economic contracting and appraisal methods have been developing

1062

C.-Y. Chang / International Journal of Project Management 31 (2013) 1057–1067

independently with no cross-fertilisation. As contracting problems may fundamentally change the value a project could yield, there should have some benefit to realise from integration of the insights from the former into the later. Generally speaking, in the investment of RE projects, investors are exposed to greater risks at the construction stage than at the operating stage. This is why refinancing might result in financial gains after construction work is completed on site. The danger of a front-loaded risk distribution in RE projects is compounded by the employment of EPC contracts in which all the responsibilities are put in the hand of a single supplier. If RE investors would like to make any post-contract changes, they have to confront severe holdup problems. Moreover, whereas EPC contracts allow project investors to transfer unwanted risks to the contractor, the limit of risk transfer is not properly accounted for in the current practise of contract design. Taking no account of risk transfer limit and behavioural uncertainty would lead to biases in appraisal and hamper a better use of FRM instruments in creating value. Inevitably, contractors have to operate under the constraint of a limited risk-bearing capacity, because a large loss might plunge him into insolvency. However, in reality, it is an extreme scenario. Before reaching insolvency, the contractor often finds it economically desirable to stop absorbing more losses beyond a point. As argued in Chang (2011), this point can be defined in terms of quasi-rent. In economics, quasi-rent refers to the return in excess of the minimum that a trader requires to continue and finish a transaction (Milgrom and Roberts, 1992). In other words, it stands for the expected reward for completing the transaction from a point in time during the contracting period onwards. Should this expected reward be completely wiped out by a realised cost shock, the contractor will find it indifferent between pulling out now and in the end of the transaction. Unless positive cost shocks are anticipated to occur, the contractor will prefer to pull out of the transaction after this point. Whether this preference will actually turn into action is an empirical question, entailing further investigation. Instead, this research argues from a normative perspective that risk-bearing capacity should be taken seriously in the determination of how much risk to transfer. In principle, the investor needs to pay premium for the risk transferred. According to the influential Principal–Agent theory, the premium that the contractor would charge can be approximately estimated by 1/2γb 2σ 2 (Gibbons, 2005; Milgrom and Roberts, 1992), where γ is the Arrow–Pratt coefficient of absolute risk aversion, b the contractor's risk-sharing rate and σ the standard deviation of the random cost risk (w). An implicit assumption of this practise is that, no matter how far the outturn cost deviates from its mean, the contractor's risk-bearing cost is the same (Chang, 2012). This assumption is flawed because the contractor has no incentive to take on the cost risk in excess of his quasi-rent. We can define a negative cost shock with the magnitude equal to the contractor's quasi-rent as the “breakup point” of the contract. After that point, the contractor will be better off terminating the contract. As a result, it is sensible to take quasi-rent as the limit of risk the contractor is willing to accept, and to assume he will walk away from the contract in the event of negative cost realisations going over the limit. Were contract

breakup to occur, the investor has to incur substantial more costs to bring the project back into track. It means the cost of risk transfer should not be treated as uniform across all cost realisations and thus using quasi-rent as a proxy for risk-bearing capacity should better reflect the “true” costs. Whilst some factors, such as trust embedded in the trading relationship, reputational concerns, and prospect of future work, could all work to delay the contractor's opt-out, it should be noted that none of them come without a cost. For example, premier contractors may put long-term reputation ahead of short-term profit, but they are more likely than ordinary contractors to charge a reputational premium in their bids. It is therefore advisable to estimate the effect of these factors on risk-bearing capacity separately. One more point merits discussion. It is true that, under contractual obligations, the contractor may not be able to back out of the contract without incurring heavy penalties, but it does not mean the breakup potential can be ignored. The presence of negative quasi-rent could greatly intensify the contractor's opportunistic tendency and put the client under the serious hold-up threat. It is the issue that the next section will turn to. 3.2. Incorporating the effect of behavioural uncertainty The hold-up problem caused by opportunistic motives is “a fundamental determinant of contractual and organisational structure” (Rogerson, 1992). The seriousness of this problem depends on how much more loss one party may suffer than the other, should the investments sunk in the transaction have to be re-deployed to somewhere else. Where investments are specific to the transaction, the investing party will be vulnerable to rent redistribution in renegotiations. Unlike in manufacturing contexts where asset specificity is the main cause of having the investing party locked in, the project investor's vulnerability arises from her interdependent relationship with the contractor in the contracting process. Since work starts on site, the investor will see her financial capital gradually turned into physical assets. The market value of the work in progress is significantly lower than its building cost, so the investor would rather give away some ground in renegotiations than abandon the project on the half way. However, in most cases, it is less costly to replace the contractor than sell the project altogether, so the extra costs involved in the replacement of the original contractor is a proxy for the investor's quasi-rent (Ive and Chang, 2007). This cost indicates the maximum ground the investor is willing to concede in renegotiations, because any loss over this cost will make replacement of the contractor a cheaper option (Chang, 2011). In the similar vein, there are two “stoppers” that would prevent the contractor from pulling out of the contract. First, the expectation of having a profit to realise upon completion would give the contractor a strong drive to complete the whole project. The second stopper is associated with the value fall of fixed assets if deployed in alternative uses. In current construction practise, it would be fairly modest because construction projects do not normally require heavy physical investments and most of the machines and equipment are rented from plant hire companies (Ive and Chang, 2007).

C.-Y. Chang / International Journal of Project Management 31 (2013) 1057–1067

Generally, the investor has much more to lose in the event of project disruption, so her bargaining position in negotiations is fairly weak. Unfortunately, renegotiations arising from change orders are rampant in the construction industry (Bajari and Tadelis, 2001). In the absence of competition pressure, the investor will be subject to overcharge and face expected profit erosion (Winch, 1989). From the perspective of transaction cost economics, the investor would fight bitterly to protect her own interest, leading to both parties incurring additional transaction costs (Williamson, 1985, 1996). To minimise these costs, the governance structure should be chosen to align with transaction attributes. By contract, in the property rights theory of the firm, both parties are assumed to divide quasi-rent frictionlessly through Nash bargaining (i.e., 50:50 split) (Hart, 1995). The ex post exploitation of the investing party's vulnerability would blunt his/her incentive to make transaction-specific, but productivity-enhancing, investments, causing so-called underinvestment problems. A similar problem exists in construction procurement as the investor's quasi-rent is also subject to redistribution in the negotiations for the price of change orders. However, there is one issue that the literature has ignored: ex post rent appropriation has a benefit of increasing the contractor's risk-bearing capacity. Change orders sometimes cannot be avoided entirely for some reasons (such as short of time in completing the design), so a better use of this “benefit” may result in some efficiency savings. It can serve as an extra buffer to help cut back the investor's cost of financial protection by, say, reducing the insurance cover needed. These arguments can be expressed in a formal way. Suppose a project faces a normal random disturbance (w) with mean 0 and variance σ 2, i.e., w ~ N(0, σ 2). The standard probability density function and standard cumulative density function of w is f(w) and F(w) respectively. This shock can be interpreted as the combined effect of all risk sources facing the project. How likely would the project withstand this shock? As discussed before, the risk-bearing capacity of a project depends on whether change orders will be issued. In the case of no change orders (with probability πn), the contractor will absorb

1063

the shock until his quasi rent (QR c,t) is completely wiped out. The likelihood of contract breakup in the absence of change orders (An) is   −Q Rc;t An ¼ F : ð2Þ σ If change orders occur, the client is vulnerable in renegotiations. According to Chang (2013), the part that the client's quasi rent would be bargained away is equal to half of the difference in the client's and the contractor's quasi rent (i.e., 1/2 (QR p,t − QR c,t)). This redistributed rent will increase the contractor's risk-bearing capacity to 1/2 (QR p,t + QR c,t). As a result, the breakup probability under change orders (Ac) is  p;t ! Q R þ Q Rc;t : ð2Þ Ac ¼ F − 2σ Graphically, two breakup probabilities are shown by the dotted areas in Fig. 3. The average likelihood of contract breakup (A) is therefore A ¼ π n An þ ð1−πn ÞAc :

ð3Þ

Risk-bearing capacity in the operating period can be measured on the similar basis. Once the project is in operation, capital investments are all sunk. At this time, the receipt from sales of project assets is insufficient to recoup their cost. It follows that the investor would rather withstand the impact from either revenue shortfall or operating cost overruns than sell the project outright. This potential loss can actually act as a buffer for financial shocks. 4. A RBC based approach 4.1. A preliminary model The aim of this section is to illuminate how the concept of risk-bearing capacity can be fit into the discounted cash flows Prob.

πn

f An

No

-QR c,t/σ

Will change orders be issued?

w/σ

Prob. F

Yes 1- π n

f

Ac -(QR c,t+QR c,t)/σ Prob.

Fig. 3. Relationship between risk-bearing capacity and breakup probabilities.

w/σ

1064

C.-Y. Chang / International Journal of Project Management 31 (2013) 1057–1067

method in the context of a simple project. A fuller modelling will be left to the future study. Suppose a T-year project requires an initial lump-sum investment I in year one. After the project starts to operate from year two, it will generate a cash flow stream CFi, i = 2,…,n. The joint impact of risk sources of this project is captured by the random cost shock w. The investor transfers this risk to the contractor through a fixed-cost design–build contract with a price covering both the forecast construction cost (I) and the risk premium (1/2γσ 2). Now that all risks have been offloaded to the contractor, the project cash flows should be discounted at a risk-free rate (rf). Therefore, the net present value of the project is 0

NPV ¼ present value of construction cost and contractor s profit

ð4Þ

þpresent cash flows  value of net  operating T X 1 1 CF t 2 I þ γσ ¼− þ  t : 1 þ rf 2 t¼2 1 þ r f

The project is a good investment if the net present value obtained from Eq. (4) is greater than zero. The condition for a “go” decision (NPV ≥ 0) is therefore T X 1 CF t I þ γσ 2 ≤  t−1 : 2 t¼2 1 þ r

ð5Þ

f

This calculation is deficient because it failed to take into account the possibility that a contract may not survive some realisations of the cost shock. To simplify, assume the shock only materialises once in the construction stage (i.e., year 1) and its occurrence will result in a contract breakup cost (L). The expected value of this cost is therefore L ¼ AL:

ð6Þ

Prob.

f III II

I

VI -QRc,t/σ -1/2γσ

w/σ

Fig. 4. Four cases for understanding the relationship between random cost realisation, risk premium, and risk-bearing capacity.

range of − 1/2γσ and −QR c,t / σ (Zone III in Fig. 4), the client becomes the winner as the loss turns out larger than the premium paid. However, in case of the shock exceeding the risk-bearing capacity (Zone VI in Fig. 4), the loss arising from contract breakup would backfire on the client. In the presence of this loss, the traditional NPV approach could result in an underestimation of costs. Considering both change-orders and no-change-orders cases, the probability that the NPV approach could give rise to a biased investment decision is the same as A. Like the UNEP reports, this research also maintains that the benefit of FRM instruments should be taken into account in the appraisal of RE projects. Eq. (7) can form a basis for evaluating how much each of risk management instruments should be used. Suppose there is a combination i consisting of j types of instruments (j = 1, 2,…,n). Its use can reduce the probability of contract breakup to Ac,i and An,i for the cases of change orders and no change orders, respectively, leading to a lower breakup loss (L i ).   L i ¼ π n An;i þ ð1−πn ÞAc;i L:

Eq. (6) can be accommodated into Eq. (4),

ð9Þ

NPV0 ¼ present value of construction cost; contractor0 s profit and breakup cost þ present cash flows  value of netoperating T X 1 1 2 CF t I þ rσ þ L þ ¼−  t : 1 þ rf 2 t¼2 1 þ r f

ð7Þ Now the “go” condition is changed to T X

1 CF t I þ γσ 2 þ L≤  t−1 : 2 t¼2 1 þ r f

This benefit has to be weighed up against the costs of the instruments used (a product of unit price pi,j and the cover chosen for the instrument j, ci,j). After the combination i is in place, the project NPV can be calculated by modifying Eq. (7), i.e., NPVi ¼ present value of construction cost; contractor0 s profit;

ð8Þ

It is interesting to examine the “go” conditions in Eq. (5) and Eq. (8) in probabilistic terms. The probability space can be divided into four parts according to the realised value of the cost shock w. Provided the cost shock turns out to be favourable (w ≥ 0, Zone I in Fig. 4), the contractor is the winner as the risk premium charged becomes part of his profit. If the realisation is in the range of zero and − 1/2γσ (Zone II in Fig. 4), the contractor will see part of his risk premium eaten away by the negative cost shock and, in the mean time, the client will receive some benefit from the insurance provided by the contractor (because the shock will otherwise fall on the client). When the realisation falls in the

ð10Þ

breakup cost under protection and protection cost þ present 2 value of net operating cash 3 flows n T X X 1 4 1 2 CF t I þ rσ þ L i þ ci;j pi;j 5 þ ¼−  t : 1 þ rf 2 1 þ rf t¼2 j¼1

Compared to Eq. (7), there are two differences in Eq. (10): first, breakup probabilities are altered to reflect the lower likelihood of contract failure under the protection of combination i; second, a new term is added to represent the cost of obtaining this protection. It is clear from Eq. (10) that the optimal choice of instruments involves a trade-off between the reduction in breakup losses as a result of their use and the cost of securing them. In the design of the procedure for finding the optimal combination, two

C.-Y. Chang / International Journal of Project Management 31 (2013) 1057–1067

points should be noted: first, as shown in the UNEP reports, the range of instruments available for choice is subject to the local market condition, so the effects of the possible combination of instruments need be examined individually; second, the contribution of an instrument to the increase in risk-bearing capacity is not linear. On account of these reasons, this research proposes a three-step procedure: first, a baseline NPV (NPV0) should be estimated for the base case where a fixed price contract was employed and no protections are used (Eq. (7)); second, an instrument combination i can then be added onto the project to check if the new net present value (NPVi) (Eq. (10)) is the lowest among the combinations that have been computed. Third, the process goes on until all of the feasible combinations have been checked. Whilst the proposed method is built on the assumption of single incidence of a time-invariant cost shock, it holds flexibility to fit into the framework of real options analysis and decision analysis by modelling the cost shock as a stochastic process. Allowing for multiple incidences might pose some challenges, but it is not technically insurmountable. It is a new line that merits further exploration. It is also worth noting that switching from the NPV approach to the new approach can avoid underestimating costs (and thus lead to a sounder decision), whilst maintaining practicality. The information needed for the application of Eq. (10) is no more than what has been required to collect in the current appraisal practise. Probability of requirement changes is a standard input in the value for money assessment of PFI business cases in the UK (HM Treasury, 2007). The methods for estimating the coefficient of risk aversion has been well established (Cardenas and Carpenter, 2008; Holt and Laury, 2002). Probabilistic forms of project risks can be found in the engineering risk management literature (Grimsey and Lewis, 2002; Ye and Tiong, 2000; Zhang, 2005).

4.2. A demonstration example For ease of exposition, this illustrative example is kept as simple as possible. Suppose a risk-averse investor plans to invest $100 million in the building of a 40 MW offshore wind farm. This project will take one year to construct and can operate for 19 more years. The total of fixed and variable operating and maintenance costs is $25 million per annum. In return, the investor expects to recoup $40 million per annum from sales of electricity to the grid. For simplicity, both tax and depreciation are not considered and assume that cost shock only happens once in the construction stage. The investor decides to procure the project through an EPC contract, under which all construction risks are transferred to the contractor. The joint effect of construction risks is a normal variable with mean of 0 and standard deviation of $20 million. Suppose the prospective contractor is averse to risk with a coefficient of absolute risk aversion 0.05. Given this belief, the contractor is expected to put a risk premium of $10 million (= 1/2 × 0.05 × 20 2) on top of the estimated construction cost in pricing the project. Risk-bearing is assumed to be the only source of the contractor's profit on this project, because competition will

1065

constrain his profit to be no more than a compensation for accepting the risks transferred. During the construction period, the contractor needs to make a lump-sum investment of $10 million to purchase special equipment needed in the installation of the wind farm. Thirty percent of its value will be lost if switched to the best alternative use. According to experience, there is 10% probability that the investor would make changes to output requirements after contract signature, prompting renegotiations to determine the price of the new work. The outcome of renegotiations considerably depends on the contractor's bargaining position. Under such a highly integrated procurement system as EPC, change of the contractor on the halfway would result in a sizable loss. Assume it will cost 30% of the project's capital value to resume the project. Given the current risk-free rate of 10%, we can know from Eq. (4) that the net present value of this project is $14.07 million. Yet, the contract may break up once the risk shock exceeds the risk-bearing capacity. We have two cases to consider. If no change orders are issued, the risk the contractor can bear is limited by his quasi-rent, i.e. the sum of risk premium ($10 million) and value loss of the lump-sum investment in alternative use ($3 million). This buffer can limit the breakup probability to   10 þ 3 F − ¼ 0:26: 20 Graphically, it is the area under the dotted line (An) in Fig. 5(a). In the event of change orders, the contractor would exploit the investor's vulnerability until her quasi-rent is all lost. In this case, the breakup probability becomes:   ð10 þ 3Þ þ 30 F − ¼ 0:02: 20 Suppose the investor estimates that there is 10% probability that requirements may need to be changed at post-contract stage, so the expected value of breakup costs is ð90%  0:26 þ 10%  0:02Þ  30 ¼ 7:08: Relative to the expected profit of 15 million, ignoring this cost will result in a 47.2% bias in appraisal. As reported in UNEP (2007), FRM instruments are prevalently used in RE projects. Suppose the investor decides to buy $10 million “Contractors All Risks” insurance to hedge against the adverse realisation of construction risks. The premium of this cover is $1 million. Is this cover value for money? If we only consider the expected value of the cost shock (i.e. zero), this protection appears not worthwhile. However, once its effect on breakup potential is considered, the answer will change. With the insurance cover, breakup probabilities can be lowered to 0.13 and 0 respectively in both cases. The breakup cost under insurance protection is reduced to 3.51 million, tantamount to yielding a saving in breakup costs of $3.57 million (7.08 − 3.51), so the net benefit of this protection is actually $2.57 million (= 3.57 − 1). This analysis can form a basis for the comparison of desirability

1066

C.-Y. Chang / International Journal of Project Management 31 (2013) 1057–1067

Prob.

References

F

0.65 An

w/σ

(a) the case of change orders Prob.

2.15

Ac

w/σ

(b) the case of no change orders Fig. 5. Risk-bearing capacity and contract breakup for a point in time.

of alternative protection choices. In the long run, it can even develop into a method enabling project investors to devise an optimal mix of risk management instruments to align with the risk profile of a project.

5. Conclusions This research critically reviews the latest development of appraisal techniques for RE projects with an aim to identify their limitations and the areas where they can be improved upon. It is found that there are three main advances in the field: first, the functions of FRM instruments in project evaluation were systematically explored; second, real options embedded in RE projects started to receive due treatment in project evaluation; third, the effort at incorporating the logics of real option analysis into the framework of decision analysis and system dynamics seems beneficial. However, we also find these approaches have two deficiencies: first, the hold-up threat resulting from behavioural uncertainty is ignored; second, the danger of contract breakup and its attendant costs is not accounted for in the appraisal of RE projects. This research sets out a preliminary model to illustrate how these two problems can be addressed through incorporation of the notion of risk-bearing capacity into the NPV framework. Demonstrably, there is a probability (A) that the NPV approach would result in a biased investment decision. The employment of the new approach would not only avoid underestimation of costs, but also provide an avenue to integrate the choice of FRM instruments in a coherent way.

Amram, M., Kulatilaka, N., 2000. Strategy and shareholder value creation: the real options frontier. Journal of Applied Corporate Finance 13 (2), 8–21. Bajari, P., Tadelis, S., 2001. Incentives versus transaction costs: a theory of procurement contracts. The RAND Journal of Economics 32 (3), 387–407. Black, F., Scholes, M., 1973. The pricing of options and corporate liabilities. Journal of Political Economy 81, 637–654. Bolton, P., Dewatripont, M., 2005. Contract Theory. MIT Press, Cambridge. Boomsma, T.K., Meade, N., Fleten, S.E., 2012. Renewable energy investments under different support schemes: a real options approach. European Journal of Operational Research 220 (1), 225–237. Borges, A.R., Antunes, C.H., 2003. A fuzzy multiple objective decision support model for energy-economy planning. European Journal of Operational Research 145, 304–316. Brach, M.A., 2003. Real Options in Practice. John Wiley & Sons. Brandao, L.E., Dyer, J.S., 2005. Decision analysis and real options: a discrete time approach to real option valuation. Annals of Operations Research 135, 21–39. Brandao, L.E., Dyer, J.S., Hahn, W.J., 2005. Using binomial decision trees to solve real-option valuation problems. Decision Analysis 2 (2), 69–88. Brennan, M., Schwartz, E., 1985. Evaluating natural resource investments. Journal of Business 58, 135–157. Cardenas, J.C., Carpenter, J., 2008. Behavioural development economics: lessons from field labs in the developing world. Journal of Development Studies 44, 311–338. Chang, C.Y., 2011. Towards an approach to measuring the risk-bearing capacity of construction contracts. Proceedings of the 2011 IEEE International Technology Management Conference, June 27–30, San Jose, California USA. Chang, C.Y., 2012. A critique of the principal–agent theory as applied to the design of engineering contracts. Working Paper, Bartlett School of Construction and Project Management. University College London. Chang, C.Y., 2013. Understanding the hold-up problem in the management of megaprojects: the case of the Channel Tunnel Rail Link project. International Journal of Project Management 31, 628–637. Chang, C.Y., Ive, G., 2002. Rethinking the multi-attribute utility approach based procurement route selection technique. Construction Management & Economics 20, 275–284. Clemen, R.T., 1997. Making Hard Decisions: An Introduction to Decision Analysis. Duxbury Press, Belmont. Contracting Excellence, 2004. Partnering in Practice: A case study demonstrating the why, what and how of Partnering, London. Copeland, T., Antikarov, V., 2001. Real Options — A Practitioner's Guide. Texetere. Copeland, T., Tufano, P., 2004. Real world way to manage real option. Harvard Business Review 82 (30), 90–99. Cox, J.C., Ross, S.A., Rubinstein, M., 1979. Option pricing: a simplified approach. Journal of Financial Economics 7 (229), 229–263. Dailami, M., Lipkovich, I., Van Dyck, J., 1999. INFRISK: a computer simulation approach to risk management in infrastructure project finance transactions. Policy Research Working Paper 2083. Economic Development Institute, World Bank. Dixit, A.K., Pindyck, R.S., 1994. Investment under Uncertainty. Princeton University Press. EDHEC-Risk Institute, 2010. Adoption of Green Investing by Institutional Investors: A European Survey. Falconett, I., Nagasaka, K., 2010. Comparative analysis of support mechanisms for renewable energy technologies using probability distributions. Renewable Energy 35, 1135–1144. Felder, F.A., 1996. Integrating financial theory and methods in electricity resource planning. Energy Policy 24, 149–154. Fernandes, B., Cunha, J., Ferreira, P., 2011. The use of real options approach in energy sector investments. Renewable and Sustainable Energy Reviews 15, 4491–4497. Gibbons, R., 2005. Incentives between firms (and within). Management Science 51 (1), 2–17.

C.-Y. Chang / International Journal of Project Management 31 (2013) 1057–1067 Grimsey, D., Lewis, M.K., 2002. Evaluating the risks of public private partnerships for infrastructure projects. International Journal of Project Management 20, 107–118. Haralambopoulos, D., Polatidis, H., 2003. Renewable energy projects: structuring a multi-criteria group decision-making framework. Renewable Energy 28, 961–973. Hart, O., 1995. Firms, Contracts and Financial Structure. Oxford University Press, Oxford. Hart, O., Moore, J., 2008. Contracts as reference points. Quarterly Journal of Economics 123 (1), 1–48. Holt, C.A., Laury, S.K., 2002. Risk aversion and incentive effects. The American Economic Review 92, 1644–1655. Howell, S., Stark, A., Newton, D., Paxson, D., Cavus, M., Pereira, J., Patel, K., 2001. Real Options: Evaluating Corporate Investment Opportunities in a Dynamic World. Financial Times/Prentice Hall. Huang, J., Poh, K., Ang, B., 1995. Decision analysis in energy and environmental modeling. Energy 20, 843–855. Hull, J.C., 2003. Options, Futures and Other Derivatives. Prentice Hall, New Jersey. ICE (Institute of Civil Engineers), 1998. RAMP: Risk Analysis and Management for Projects. Thomas Telford, Trowbridge. Ive, G., Chang, C.Y., 2007. The principle of inconsistent trinity in the selection of procurement systems. Construction Management and Economics 27 (7), 677–690. Kumbaroğlu, G., Madlener, R., Demirel, M., 2008. A real options evaluation model for the diffusion prospects of new renewable power generation technologies. Energy Economics 30, 1882–1908. Lee, S.C., 2011. Using real option analysis for highly uncertain technology investments: the case of wind energy technology. Renewable and Sustainable Energy Reviews 15, 4443–4450. Lee, S.C., Shih, L.H., 2010. Renewable energy policy evaluation using real option model — the case of Taiwan. Energy Economics 32, S67–S78. Lee, S.C., Shih, L.H., 2011. 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 15, 2185–2198. Luehrman, T.A., 1997. What's It Worth? A General Manager's Guide to Valuation. Harvard Business Review 132–142 (May–June). Luehrman, T.A., 1998a. Investment Opportunities as Real Options: Getting Started on the Numbers. Harvard Business Review 3–15 (July–August). Luehrman, T.A., 1998b. Strategy as a Portfolio of Real Options. Harvard Business Review 89–99 (September–October). Milgrom, P., Roberts, J., 1992. Economics, Organization & Management. Prentice Hall, New York. Mun, J., 2002. Real Options Analysis: Tools and Techniques for Valuing Strategic Investments and Decisions. John Wiley. Myers, S.C., 1984. Finance theory and financial strategy. Interfaces 14, 126–137. NAO, 2001. The Channel Tunnel Rail Link. HC302 Session 2000–2001. The Stationery Office, London. NAO, 2004. London Underground PPP: Were They Good Deals? HC 645 Session 2003–2004. The Stationery Office, London. NAO, 2006. The Termination of the PFI Contract for the National Physical Laboratory, HC 1044 Session 2005–2006. The Stationery Office, London. NAO, 2009. The Failure of Metronet, HC 512 Session 2008–2009. The Stationery Office, London. Owens, G., 2002. Best Practices Guide: Economic & Financial Evaluation of Renewable Energy Projects. U.S. Agency for International Development, Washington DC.

1067

Ozerdem, B., Ozer, S., Tosun, M., 2006. Feasibility study of wind farms: a case study for Izmir, Turkey. Journal of Wind Engineering and Industrial Aerodynamics 94, 725–743. Ritchie, B., Brindley, C., 2007. An emergent framework for supply chain risk management and performance measurement. Journal of the Operational Research Society 58, 1398–1411. Rogerson, W.P., 1992. Contractual solutions to the hold-up problem. Review of Economic Studies 59, 777–794. San Cristóbal, J., 2011. Multi-criteria decision-making in the selection of a renewable energy project in Spain: the Vikor method. Renewable Energy 36, 498–502. Simon, P., Hillson, D., Newland, K., 1997. PRAM: Project Risk Analysis and Management Guide. The Association for Project Management, Trowbridge. Siskos, J., Hubert, P., 1983. Multi-criteria analysis of the impacts of energy alternatives: a survey and a new comparative approach. European Journal of Operational Research 13, 278–299. Smith, J.E., McCardle, K.F., 1998. Valuing oil properties: integrating option pricing and decision analysis approaches. Operations Research 46 (2), 198–217. Smith, J.E., Nau, R.F., 1995. Valuing risky projects: option pricing theory and decision analysis. Management Science 41 (5), 795–816. Tan, B., Anderson Jr., E.G., Dyer, J.S., Parker, G.G., 2010. Evaluating system dynamics models of risky projects using decision trees: alternative energy projects as an illustrative example. System Dynamics Review 26 (1), 1–17. Torriti, J., 2012. Multiple-project discount rates for cost–benefit analysis in construction projects: a formal risk model for microgeneration renewable energy technologies. Construction Management and Economics 30, 739–747. Treasury, H.M., 2007. Quantitative Assessment User Guide. HMSO, London. UNEP, 2004. Scoping Study on Financial Risk Management Instruments for Renewable Energy Projects, United Nations Energy Program. UNEP, 2006. Background Study, United Nations Energy Program. UNEP, 2007. Assessment of Financial Risk Management Instruments For Renewable Energy Projects. Working Group 1 Study Report, United Nations Energy Program. UNEP, 2008. e-Learning Course on Insurance Risk Management for Renewable Energy Projects, United Nations Energy Program. Venetsanos, K., Angelopoulou, P., Tsoutsos, T., 2002. Renewable energy sources project appraisal under uncertainty: the case of wind energy exploitation within a changing energy market environment. Energy Policy 30, 293–307. WEF, 2011. Green investing 2011: reducing the cost of financing (2011). World Economic Forum. Williamson, O.E., 1985. The Economic Institutions of Capitalism. The Free Press, New York. Williamson, O.E., 1996. The Mechanics of Governance. Oxford University Press, Oxford. Winch, G., 1989. The construction firm and the construction project: a transaction cost approach. Construction Management and Economics 7, 331–345. Ye, S., Tiong, R.L.K., 2000. NPV-at-risk method in infrastructure project investment evaluation. Journal of Construction Engineering and Management 126, 227–233. Zhang, X., 2005. Financial viability analysis and capital structure optimization in privatized public infrastructure projects. Journal of Construction Engineering and Management 131, 656–668. Zhou, P., Ang, B., Poh, K., 2006. Decision analysis in energy and environmental modeling: an update. Energy 31, 2604–2622.