Household willingness to pay for expanding fuel cell power generation in Korea: A view from CO2 emissions reduction

Renewable and Sustainable Energy Reviews 81 (2018) 242–249

Contents lists available at ScienceDirect

Renewable and Sustainable Energy Reviews journal homepage: www.elsevier.com/locate/rser

Household willingness to pay for expanding fuel cell power generation in Korea: A view from CO2 emissions reduction Seul-Ye Lim, Hyo-Jin Kim, Seung-Hoon Yoo

MARK

⁎

Department of Energy Policy, Graduate School of Energy & Environment, Seoul National University of Science & Technology, 232 Gongreung-Ro, Nowon-Gu, Seoul 01811, Republic of Korea

A R T I C L E I N F O

A BS T RAC T

Keywords: Fuel cell power generation Contingent valuation Spike model Willingness to pay

The fuel cell (FC) is accepted as the state-of-the-art instrument for reducing CO2 emissions in the field of power generation in Korea. Therefore, the Korean government has planned to expand FC power generation from 565 GWh in 2013 to 13,449 GWh by 2027. This article aims to assess the public value or acceptability of this expansion in terms of reducing CO2 emissions. To this end, we derive the public's additional willingness to pay (WTP) for the expansion using a contingent valuation survey of 1000 Korean households. For the purpose of mitigating the response effect in eliciting their WTP and increasing statistical efficiency in analyzing the WTP data, we employ a one-and-one-half-bounded dichotomous choice question format. Furthermore, we use the spike model so as to model zero WTP responses. The mean additional monthly WTP for the expansion is computed to be KRW 1407 (USD 1.37) per household. Expanding the value to the national population gives us KRW 310.5 billion (USD 304.4 million) per year. We can conclude that Korean households are ready to shoulder some of the financial burden of expanding FC power generation to reduce CO2 emissions.

1. Introduction Korea has an export-led economy that relies on manufacturing, which emits more greenhouse gases than the service sector. The nation was ranked as the seventh largest emitter of carbon dioxide (CO2) in 2014, following China, the USA, India, Russia, Japan, and Germany. At the Paris Climate Change Conference in December 2015, Korea, which is under domestic and international pressure to lessen CO2, announced its commitment to reduce its CO2 emissions by 37% from the businessas-usual level by 2030. Of this reduction target, 25.7% applies to the reduction of CO2 emissions in the country and 11.3% depends on international CO2 emissions trading systems. The principal emitter of CO2 in Korea is the manufacturing sector. For instance, the manufacturing sector consumed 62.3% of final energy and accounted for 38.4% of GDP as of 2013 [1,2]. The largest consumer of energy in the manufacturing sector is heavy and chemical industries and more than half of their production is not consumed in Korea and exported abroad. In addition, Korea currently is the sixth-largest steel producer in the world and a large portion of its steel (39% in 2011) is produced by electric arc furnaces [3]. For the sake of economic growth, the Korean government has kept industrial electricity price quite low by spewing CO2 from coal-fired power plants and has not regulated CO2 emissions of the sector.

⁎

In short, the competitiveness of the Korean manufacturing results from low electricity price and high CO2 emissions. The sector should invest or spend lots of money additionally to significantly abate CO2 emissions target. Mitigating CO2 emissions in the manufacturing sector is quite a difficult prospect as it will certainly have significant negative impacts on economic growth. Moreover, the building and transportation sectors do not have a room for reducing CO2 emissions because sufficient investments on and possible measures for the reduction have been already made. Therefore, the Korean government has placed relatively loose duty to reduce CO2 emissions on the manufacturing, building, and transportation sectors. To achieve the CO2 mitigation target, the Korean government has focused on the generation sector and planed to replace fossil fuels with energy sources that emit low level of carbon. Policy makers are currently addressing the potential effectiveness of regulations and other measures for reducing CO2 emissions in generation sector. For example, the government has introduced a renewable portfolio standard that obligates the power generators to supply a portion of electricity from new and renewable energy sources since 2012. The portion was 3.0% in 2012 and set to increase to 12.6% in 2027 [4,5]. Electric power producers are obligated to supply more electricity from new and renewable sources than it has done hitherto and assist them in doing so.

Corresponding author. E-mail addresses: [email protected] (S.-Y. Lim), [email protected] (H.-J. Kim), [email protected] (S.-H. Yoo).

http://dx.doi.org/10.1016/j.rser.2017.07.038 Received 1 October 2016; Received in revised form 24 February 2017; Accepted 9 July 2017 Available online 03 October 2017 1364-0321/ © 2017 Elsevier Ltd. All rights reserved.

Renewable and Sustainable Energy Reviews 81 (2018) 242–249

S.-Y. Lim et al.

for the expansion through an increase in their current electricity bill using a contingent valuation (CV) survey. Moreover, WTP can be taken as indicative of the public preference or acceptability of FC power generation policy with a view to reducing CO2 emissions. In connection with electricity bill, Korean customers have no choice about a power company in electricity market because of existence of only one electricity seller in market. However, the Korean government can expand FC power generation using regulation and/or incentive schemes on electric power companies. This inevitably causes an increase in the production costs of electricity and a rise in electricity price. These points were explicitly explained to the respondents in the CV survey. In this regard, the household WTP implies the degree of the public pereference or acceptability of the FC power generation expansion although the household has no choice about a power company. Therefore, the prime objective of our paper is to measure the household WTP for implementing the policy of expanding FC power generation from 565 GWh in 2013 to 13,449 GWh by 2027. This aim is met using a stated preference technique based on a survey of consumers, that is, the CV. The remainder of the paper comprises four sections. The methodology employed in this study is described in Section 2. The modeling of WTP responses is addressed in Section 3. The results are explained and discussed in Section 4. The paper is concluded in Section 5.

In this regard, photovoltaic power and wind power generation facilities have drastically increased in Korea. However, back-up fossilfueled power plants with the same generation capacity are needed to safeguard the supply of electricity because photovoltaic power is intermittent. In contrast, fuel cell (FC) generation, a representative new energy, is not intermittent and attracting attention as the state-ofthe-art instrument for reducing CO2 emissions in the field of power generation. FC generation using hydrogen with oxygen provides clean, efficient, and reliable electricity. Although some stationary FCs require the use of fossil fuels, such as natural gas or hydrocarbons, as a hydrogen feedstock, they produce far fewer CO2 emissions than conventional fossil-fueled power plants. Furthermore, active progress is being made in technological developments to provide FC power generation at a lower cost and with lower space requirements [6–9]. FC has three other promising merits in Korea. First, FC can reduce the severe energy dependence on foreign countries and increase energy supply security [10–13]. Energy supply independence is quite an important task for Korea because the country suffers a scarcity of energy and is one of the heavy energy importer countries. For example, the country imported 95.2% of the energy it consumes from abroad as of 2014. However, the hydrogen needed for FC power generation can be obtained from domestic energy sources. Second, FC power among new and renewable energy can generate near electricity-consuming areas, and do not require large-scale or long-distance power transmission facilities. All the nuclear power and coal-fired plants, presently a base-loaded and dominant power generation source in Korea, are located around coastal areas in order to obtain sea water for cooling, and they are quite remote from metropolitan areas for safety reasons. Consequently, the nuclear power and coalfired plants inevitably require large-scale and long-distance power transmission facilities as they are located far from the consumers. The rights of way for high voltage lines often run through high-value land and reduce the property values there. This tends to result in a substantial amount of social costs (e.g., [14]). Third, energy efficiency, defined as the ratio of practically effective energy compared to the energy consumed by energy converters, such as engines, green plants, turbines, etc., is relatively high for FCs. In general, their energy efficiency is known to be high, ranging from 40% to 60% [15]. Indeed, if the FC uses the heat generated from the reaction process, up to 80% of the fuel can be changed into energy. For these reasons, a number of countries have started and will continue to introduce the use of FC in various fields. For example, there are the installations of an FC-based combined cooling, heating, and power system in Sweden [16], the FC system for commercial buildings’ thermal and electric energy storage in the US [17], microbial FC generation in China [18], and FC power generation and its storage facilities in Pakistan [19]. The Korean government also plans to expand FC power generation [3]. The amount generated from FC was 565 GWh in 2013, but it will reach 13,449 GWh by 2027 if the plan is successfully completed. This means increasing the ratio of FC power generation from 0.2% in 2013 to 2.0% by 2027. The main instruments for achieving this goal include governmental support for the development of FC technology and the new introduction of a standard mandate for the renewable portfolio using FC. Our study attempted to obtain information on the household preference or acceptability for implementing FC power generation expansion policy. Clearly, the public preference for expanding FC power generation to achieve the aforementioned merits can be employed as a proper and important reference for further discussion of the expansion of FC and decisions on establishing power transmission facilities and power generation. Furthermore, the public acceptability of FC power generation expansion policy is quite important to making any decisions when the government is considering the introduction of green pricing demanding electricity price rise following FC power generation expansion. Public preference or acceptability can be investigated by assessing people's additional willingness to pay (WTP)

2. Methodology 2.1. Method of assessing the additional WTP for FCs: the CV approach In the context of economics, a WTP of consuming a commodity can be determined by computing the area below the demand curve for the commodity. The area is precisely the consumer's WTP for the commodity. Thus, the first step in evaluating the WTP is to estimate the demand function for the commodity and the next step is to calculate the area under the demand function. However, if the commodity is not traded on the market, in other words, if it is a non-market good, it is difficult to estimate the demand function. In this case, the area can be directly computed by using a stated preference technique such as the CV method rather than estimating the demand function and then calculating the area under it. In our study, the demand function is defined for environmental quality improved by the expansion of FC power generation. The expansion is a case for which directly calculating the area under the demand function is an appropriate strategy. Therefore, we estimate the household WTP for implementing FC power generation expansion policy with a view to reducing CO2 emissions using the CV technique. Some evidences that the WTP from reducing CO2 emissions is related to people's WTP in Korea are found in the literature. For example, Yoo et al. [20] and Lee et al. [21] show that the Korean households have positive WTP for reducing GHG emissions. Moreover, the Korean people are willing to pay a premium for renewable energies (e.g., [22,23]). Thus, our approach is consistent with the practice of former studies that have measured WTP in relation to expanding the use of new or renewable energies (e.g., [24–30]). It asks randomly chosen respondents a question concerning their WTP for enhancing the amount of electricity consumed from FC generation using a wellstructured survey [31,32]. The CV technique is the one most widely applied in the literature to obtain information on people's WTP for consuming or obtaining nonmarket goods [33]. There are no restrictions on the objects that can be valued using the CV method. In particular, it is more useful than other methods because it can capture the non-use or existence value of goods, which cannot be measured through a market mechanism. Nonmarket goods include environmental goods or public goods, such as FC power generation. Thus, as explained earlier, this study seeks to use the CV approach to assess the additional WTP for the expansion. 243

Renewable and Sustainable Energy Reviews 81 (2018) 242–249

S.-Y. Lim et al.

also checked the consistency of the respondent's answers by asking several questions again. Perhaps remarkably, respondents in Korea understood the CV questions easily with the help of the interviewer. In the process of verification 400 observations from the original total of 1400 interviewed were removed from the sample. For some observations, we could find no one by the given name at the telephone number given by the respondent. For some questionnaires, the answers given over the phone were inconsistent with the answers given in the interview. For some survey results, there were skipped variables that cannot be made up over the phone. Some respondents frankly confessed that they did not pay attention to the CV survey. Some observations were evaluated to be of poor quality by the interviewers. Finally, we obtained 1000 useable observations.

Some may doubt the practicality and usefulness of the CV method because it gathers information from a survey of respondents. In this regard, the blue-ribbon National Oceanic and Atmospheric Administration (NOAA) Panel came to the influential conclusion that the CV method can produce reliable quantitative information that can be used in decision making by both public administrations and judicial bodies, provided that several guidelines proposed by the NOAA Panel are observed [34]. Moreover, following the guidelines can secure the validity and accuracy of the CV method. For example, the goods of concern should be familiar to the public, the CV survey should be administered through face-to-face interviews by professionally-trained interviewers rather than through telephone or mail interviews, a suitable payment vehicle should be adopted and presented to the respondents, and the substitutes for the goods should be explained to the respondents in the survey. The conditions are met in our study, as discussed in detail below.

2.3. Two CV survey design issues Concerning the CV survey design, two issues and how to deal with them in our study should be addressed here. The first issue is whether the respondents could isolate the CO2 emission reduction only among a number of merits of FC generation. The respondents may not regard CO2 emission as the “only” attribute of FC power generation but consider other attributes of FC power during their value judgments. For example, some merits such as the reduction of the air pollutants emissions and/or a decentralized generation might be reflected in stating their WTP for FC power generation. This may cause an overestimation of the WTP. In this regard, we tried to avoid any overestimation or embedding problem in two ways. First, we gave the interviewers sufficient information about the purposes and background of the survey, and instructed them how to answer the questions that might be raised by the interviewees in a perspective of FC power generation. The supervisors affiliated with the survey company trained the interviewers to implement the survey as persuasively and effectively as possible. Second, in the CV survey, we explicitly explained that the focus of the survey is just the role of FC power generation from a standpoint of mitigating CO2 emissions. The respondents were told that they should ignore other attributes of FC power generation. That is, the following statement was presented to the respondents. “Please bear in mind that you should ignore other merits of FC power generation such as the reduction of the air pollutants emissions and/or a decentralized generation source that does not demand long and large-scale transmission facilities when stating your household's willingness to pay for FC power generation expansion.” We think that we successfully excluded other attributes from the respondents’ judgments. Moreover, judging from the interviewers’ comments, the respondents gave their responses after well understanding the background information conveyed to them in the survey. Thus, we believe that any overestimation or embedding problem does not appear in our study. The second issue is whether the respondents could consider the FC generation only among various new and renewable energy sources. The focus of our study is the FC generation's role in the CO2 emission reduction. Thus, the respondents should not consider other new and renewable energy sources. However, the respondents might consider other new and renewable energy sources in our CV survey emphasizing CO2 emission reduction because increasing the use of other renewable energy such as wind power and photovoltaic power can also abate CO2 emissions. As explained above, Korea Ministry of Trade, Industry, and Energy [3] announced that the Korean government will implement a policy of increasing the share of new and renewable energy in total generation from 3.0% of in 2013 to 12.6% in 2027. The FC power generation is a realistic and effective method of electricity generation as a new and renewable energy source in Korea. Thus, the new and renewable energy expansion policy includes increasing the share of the FC generation in new and renewable energy from 3.6% in 2013 and 14.9% in 2027. Considering that the main generation source in Korea is coal-fired

2.2. CV survey method We commissioned a professional survey firm to arrange the CV field survey. The firm that has a number of experiences of conducting CV surveys drew a stratified random sample during July 2014 to obtain information on the households’ WTP for FC power generation and their socioeconomic characteristics. According to ‘2010 National Population and Housing Census’ implemented by the Korea National Statistical Office [35], there were 18,457,628 households and sixteen provinces in Korea. In order to draw a random sample of this population, sampling was conducted by the professional polling firm. Our sample of 1400 households was allocated to the provinces in proportion to each province's population characteristics, such as age, income, and gender, resulting in 18 to 328 households being assigned to each province. A CV survey can be conducted using face-to-face in-person, telephone, or mail interviews. The response rate to a mail survey is usually quite low, and a telephone survey can present only a limited amount of information to respondents. We wished to convey a large amount of explanatory information on FC power generation expansion, and to provide visual cards to describe the situation with and without the project, as well as outlining the expected effects of the project to the respondents in the CV survey. Thus, we used face-to-face interviews. We gave the interviewers sufficient information concerning the purposes and background of the CV survey, and instructed them on how to answer the questions that might be raised by the interviewees. Moreover, the supervisors affiliated with the survey company trained the interviewers in implementing the CV survey as persuasively and effectively as possible. To obtain reliable and responsible information on decision making by respondents, heads of households or homemakers aged 20–65 years old were selected and interviewed in the CV survey. The survey instrument consisted of three parts. The first was an introductory section explaining the general background information concerning FC power generation to the respondents and then asking them about their perceptions of such power generation. The scenario in which the goods to be valued will be provided to the public should be explained clearly. The second part included questions concerning their WTP for expanding FC power generation by 2027. These questions should be presented in a context that ensures that the WTP questions are plausible, understandable, and meaningful. The final part contained questions related to the households’ socioeconomic variables. The post-interview follow-up telephone check was done to reduce the number of skipped questions and to verify the results of the survey, both of which tend to increase the reliability of our data. Besides obtaining answers for the skipped variables, a total of 1400 observations were verified that the CV survey was properly conducted. We asked by phone whether the interviewer performed his/her job properly, whether the interviewers used the visual aids properly, and whether the respondents sufficiently understood the CV questions. We 244

Renewable and Sustainable Energy Reviews 81 (2018) 242–249

S.-Y. Lim et al.

vehicle, and there should be a clear connection between the goods to be valued and the payment vehicle. For this reason, the electricity bill, which is familiar to most respondents and clearly related to actual expenditure, is appropriate for the payment vehicle [33,42,43]. Thus, monthly electricity bills are employed as the payment vehicle in this study. The WTP question read as follows: “Is your household willing to pay a specified bid for expanding FC power generation through an increase in monthly electricity bills, supposing that the expansion will certainly be guaranteed?” Moreover, some additional statements concerning payment were provided in the questionnaire. For example, the respondents were told: “If a majority of respondents refuse to pay the cost involved in the expansion of FC power generation, the expansion cannot be implemented. However, if a majority of respondents accept the payment, the expansion can be implemented. Please bear in mind that your household's income is constrained and that there are various expenditures in your household.”

generation, the expansion of new and renewable energy indicates the reduction of the coal-fired generation and CO2 emissions. We expect that the FC generation expansion policy will mitigate 4.2 million tones of CO2 as of 2027. This value amounts to 0.52% of total CO2 emissions in 2027. These points were clearly conveyed to the respondents and it was stressed during the CV survey that the main focus was FC power generation among new and renewable energies and they should not consider other renewable energy technologies such as solar photovoltaic, wind, and so on in responding to the WTP questions. Therefore, we and the interviewers think that the respondents successfully assess the role of FC generation in the reduction of CO2 emissions not considering other new and renewable energy sources. 2.4. Method of WTP elicitation Our study used a close-ended question format, which is preferred in the literature to open-ended questions because a respondent is likely to show strategic behavior and have difficulty in giving a WTP response when an open-ended question is asked. In addition, the blue-ribbon NOAA panel's report [34] supports the use of close-ended questions rather than open-ended questions. In particular, single-bounded (SB) dichotomous choice (DC) questions or double-bounded (DB) DC questions tend to be used. An SBDC question is a one-time DC question [36]. In contrast, a DBDC question employs two times DC questions. A respondent who states “yes” to a first bid is then asked a follow-up question concerning whether he or she would pay a second, higher bid. A respondent who answers “no” to the first bid is asked a follow-up question concerning whether he or she would pay a second, lower bid. The DBDC question format can significantly increase statistical efficiency compared to the SBDC question format [37]. However, the DBDC question format also augments response bias when compared to the SBDC question format (e.g., [38–40]). Thus, both SBDC and DBDC questions suffer from low statistical efficiency and high response bias, respectively. As an alternative to these, Cooper et al. [41] suggested the one-and-one-half bound (OOHB) DC question format. The statistical efficiency of the OOHB DC question format is similar to that of the DBDC question format and the consistency of the OOHB DC question format is close to that of the SBDC question format. The OOHB DC question format is based on a set of two bids. In an OOHB DC question, the interviewer randomly selects one of two bids, which is then given to the respondent. If the bid selected is the lower and the respondent's response is “yes,” the remaining higher bid is also presented to the respondent. If the selected bid is the lower and the respondent's response is “no,” a follow-up question is not needed. If the selected bid is the higher and the respondent's response is “yes,” a follow-up question is not needed. If the selected bid is the higher and the respondent's response is “no,” the residual lower bid is then presented to the respondent. We used seven sets of two bids, determined through a pretest on a focus group (30 persons). The list of sets used in this study is as follows: (1000, 3000); (2000, 4000); (3000, 6000); (4000, 8000); (6000, 10,000); (8000, 12,000); (10,000, 15,000). The figures given are in Korean won (KRW); the first element of each set is the lower bid and the second element is the higher bid. At the time of the survey, USD 1 was approximately equal to KRW 1020.

3. Modeling of WTP responses 3.1. Basic WTP model There are two approaches to modeling WTP responses gathered from a DCCV survey: the utility difference approach suggested by Hanemann [36] and the WTP function approach proposed by Cameron and James [44]. The first specifies the difference in utility using a random utility maximization model, while the second specifies the WTP responses directly. As pointed out by McConnell [45], the choice of which to use is not an issue of right and wrong, but depends on a researcher's preference, because the two approaches are equal in the context of economics. The literature shows that the first has been more frequently applied than the second. Thus, we adopt the utility difference approach in our study. The ratios of “yes” responses to each given bid form the basic input when applying this approach. Through the process of utility maximization under income constraints, a DC response concerning whether to pay a specified amount to attain a given environmental improvement or the provision of a public good is derived for each respondent. The independent variables of the utility function, U , include the respondent's income, socioeconomic characteristics, and perceptions of the good to be valued and its provision state. The provision state of the goods to be valued is denoted by S . The value of S is one in the case that the good is provided and zero otherwise. The respondent's income and the other factors that affect the respondent's utility are denoted by M and T , respectively. Thus, the utility function is defined as:

U = V (S , M ; T ) + ω

(1)

where V is the indirect utility function that we can obtain by inserting the solution to the utility maximization problem in the objective function of the respondent's utility, ω is a random component of the utility, and the ω ’s are independent and identically distributed random variables with zero means. The respondent will maximize his/her utility by showing that he/ she is willing to pay a presented bid, B , to obtain the goods to be valued if:

V (1, M − B; T ) + ω1 ≥ V (0, M ; T ) + ω0

(2)

Re-arranging Eq. (2) produces:

2.5. Payment vehicle

V (1, M − B; T ) − V (0, M ; T ) ≥ ω0 − ω1

A respondent may be embarrassed when asked directly for his/her WTP. Introducing in the questionnaire a medium through which the amount would be paid helps the respondent to reveal his/her true WTP. We usually call the medium the payment vehicle. Payment vehicles found in the literature include taxes, funds, donations, and expenditure. The respondents should feel at home with the payment

(3)

The left-hand side of Eq. (3) is the utility difference, defined as ΔV , and is the systematic and deterministic part, while the right-hand side is the non-systematic and random part. Let ω0 − ω1 be θ and Hθ (⋅) be the cumulative distribution function (cdf) of θ . Using Eq. (3), we can express the probability of obtaining an answer “yes” to a given bid as: 245

Renewable and Sustainable Energy Reviews 81 (2018) 242–249

S.-Y. Lim et al.

Pr{response is "yes"} = Pr{ΔV (B ) ≥ θ} = Hθ[ΔV (B )]

(4)

DBDC model. This is why we use an OOHB DC model in this study instead of an SBDC or DBDC model. The following OOHB DC model is based on Cooper et al.’s [41] suggestion. We have J observations to be analyzed. A bid, Bj , is given to respondent j for . During the CV survey, two bids, BjL and BjU , where BjL < BjU , are presented to each respondent j. Around half of the respondents are provided with BjL as the first bid. U Bj is supplied as the second bid if the answer is “yes.” In this case, there are two outcomes, “yes–yes” ( X > BjU ) and “yes–no” (BjL < X < BjU ). If the respondent answers “no” to the first bid, BjL , the outcome is “no” ( X < BjL ). The remaining respondents are presented with BjU as the first bid. If the answer is “no,” the second bid, BjL , is then supplied to the respondent and the possible outcomes are “no– yes” (BjL < X < BjU ) and “no–no” ( X < BjL ). If the respondent's response is “yes” (X > BjU ), no further bid is required. Therefore, for the six results we can introduce six binary variables, I jYY , I jYN , I jN , I jNY , I jNN , and I jY . The value of each binary variable is 1 if the respondent's response corresponds to its superscript and zero otherwise. For example, I jYY takes the value 1 if the respondent j reports “yes–yes”, and zero otherwise.

From a different perspective, we can introduce WTP, X , as a random variable in the description of the probability of responding “yes” to a presented bid, as follows.

Pr{response is "yes"} = Pr{X ≥ B} ≡ 1 − FX (B )

(5)

where FX (⋅) is the cdf of X . Comparing Eq. (4) and Eq. (5) yields:

1 − FX (B ) = Hθ[ΔV (B )]

(6)

Therefore, there is no need to assume the functional form of Hθ (⋅); all we have to do is to assume the functional form of FX (⋅) and estimate its parameters. Usually, we assume that ΔV = α − βB , where α and β are parameters to be estimated. 3.2. Model for addressing zero WTP responses: spike model Some people can have an interest in the goods to be valued, but others may be totally indifferent to or place no value on the goods. In this case, the proportion of zero WTP responses in the CV survey may be high. Researchers should pay close attention to how they deal with the WTP responses with zero observations. For this purpose, the spike model proposed by Kriström [46] is quite useful. The spike model was originally developed for SBDC CV data and adjusted for DBDC CV data by Yoo and Kwak [47]. To the best of the authors’ knowledge, a combination of the OOHB DC CV model and the spike model was suggested in Lee and Yoo [48]. Some studies applying CV utilized the OOHB DC spike model [e.g., 30,32,33,43,49,50]. This study will apply the OOHB DC spike model to model the zero WTP responses. The spike model specifies the probability of zero WTP responses as a spike at zero in the distribution of WTP, and thus enables us to analyze both zero point and positive interval WTP data in a univariate setting. In the spike model, FX (B; λ ) has the functional form:

⎧[1 + exp(α − βB )]−1 ifB > 0 ⎪ FX (B; λ ) = ⎨[1 + exp(α )]−1 ifB = 0 ⎪ ⎩0 ifB < 0

3.4. The OOHB DC spike model In order to identify zero WTP observations, we asked the respondents who answered “no” when the first presented bid was Bj L , or a “no–no” response when the first presented bid was BjU , an additional follow-up question that can tell true zero WTP from positive WTP less than BjL . Consequently, we can formulate one more binary variables, I jP , the value of which is 1 if the j th respondent's WTP is positive and zero otherwise [22,51–64]. The log-likelihood function of the OOHB DC spike model is:

(7)

J

lnL = ∑ j =1 {(I jYY + I jY )ln [1 − FX (BjU ; λ )]

As explained earlier, the spike is defined as the probability of the respondent's WTP being zero. Thus, the spike is computed as [1 + exp(α )]−1. Some covariates, such as the respondent's household income, can be incorporated in the spike model. A common method for doing this is to make the covariates penetrate into α in Eq. (7). That is, α is simply changed into α + z′δ , where z is a vector of the covariates and δ is a vector of the corresponding parameters to be estimated.

+ (I jYN + I jNY )ln [FX (BjU ; λ ) − FX (BjL ; λ )] + I jP(I jN + I jNN )ln [FX (BjL ; λ ) − FX (0; λ )] + (1 − I jP )(I jN + I jNN )ln FX (0; λ )}

(8)

Using Eq. (7), the mean of the WTP can be estimated as:

E (X ) =

3.3. The OOHB DC model

∫0

∞

0

[1 − FX (B; α , β )]dB −

∫−∞ FX (B; α, β )dB (9)

= (1/ β )ln [1 + exp(α )]

As explained above, an OOHB DC model produces higher statistical efficiency than an SBDC model and yields greater consistency than a Table 1 Distribution of responses by bid amount. Bid amounta

1000/3000 2000/4000 3000/6000 4000/8000 6000/10,000 8000/12,000 10,000/15,000 Totals

Lower bid presented as the first bid (%)

Sample sizeb

Upper bid presented as the first bid (%)

yes-yes

yes-no

no-yes

no-no

yes

no-yes

no-no-yes

no-no-no

3 (2.1) 3 (2.1) 6 (4.2) 3 (2.1) 3 (2.1) 0 (0.0) 3 (2.1) 21 (2.1)

24 (16.8) 13 (9.1) 12 (8.4) 5 (3.5) 2 (1.4) 3 (2.1) 0 (0.0) 59 (5.9)

5 (3.5) 12 (8.4) 12 (8.4) 21 (14.7) 20 (14.1) 23 (16.2) 25 (17.4) 118 (11.8)

40 (28.0) 43 (30.1) 41 (28.7) 43 (30.1) 46 (32.4) 45 (31.7) 44 (30.6) 302 (30.2)

9 (6.3) 4 (2.8) 9 (6.3) 0 (0.0) 2 (1.4) 0 (0.0) 2 (1.4) 26 (2.6)

13 (9.1) 7 (4.9) 8(5.6) 8 (5.6) 4 (2.8) 6 (4.2) 3 (2.1) 49 (4.9)

6 (4.2) 14 (9.8) 13 (9.1) 14 (9.8) 22 (15.5) 25 (17.6) 21 (14.6) 115 (11.5)

43 (30.1) 47 (32.9) 42 (29.4) 49 (34.3) 43 (30.3) 40 (28.2) 46 (31.9) 310 (31.0)

Notes: a Unit is KRW (USD 1 was approximately equal to KRW 1020 at the time of the survey). b Numbers in parentheses beside the number of responses represent the percentage of the sample size.

246

143 (100.0) 143 (100.0) 143 (100.0) 143 (100.0) 142 (100.0) 143 (100.0) 144 (100.0) 1000 (100.0)

Renewable and Sustainable Energy Reviews 81 (2018) 242–249

S.-Y. Lim et al.

4. Results

Table 3 Definitions and sample statistics of variables.

4.1. Data As explained above, we obtained 1000 useable observations. Table 1 describes the distribution of responses by each bid amount. Each set of bids was allocated to a similar number of respondents, as is shown in the last column of Table 1. As explained above, the “no” responses to the WTP question when the lower bid is presented as first bid are made up of zero WTP and positive WTP less than the lower bid. Thus, a “no” response in the lower bid presented as first bid case should be further investigated to identify which it belongs to of zero WTP and positive WTP. For this purpose, we asked a further question to identify zero WTP of the respondents who showed “no” responses when the lower bid was presented as the first bid. “no-no” and “no-yes” responses indicate zero WTP and positive WTP, respectively. The separately identified information is utilized in estimating the spike model. Both “no–no” responses when the lower bid was presented as the first bid and “no–no–no” responses when the upper bid was presented as the first bid indicate the WTP responses of zero. A total of 612 households (61.2%) revealed zero WTP for implementing the FC generation expansion policy. This implies that the use of a spike model to address WTP responses of zero is a suitable approach in our study. Moreover, a WTP response of zero is consistent with the microeconomic theory that non-consumption can be obtained as a corner solution to a utility maximization under income constraints. Overall, the proportion of “yes” responses in the responses to a given bid declines as the magnitude of the bid increases. For instance, when the lower bid was presented as the first bid, that is, when the bids moved “from lower bid to upper bid”, 27 respondents (18.9%) accepted paying KRW 1000 (USD 0.98), while just 3 respondents (2.1%) agreed to a payment of KRW 10,000 (USD 9.80).

−0.469 (−7.21)# −0.346 (−21.12)# 0.615 (39.91)# KRW 1402 (USD 1.37) 15.91# KRW 1247 to 1587 (USD 1.22–1.56) KRW 1204 to 1652 (USD 1.18–1.62) 1000 −1000.55 579.55 (0.000)

Standard deviation

Gender

Gender of the respondent (0=female; 1=male) Respondent's educational level in years Household's monthly income before tax deduction (unit: KRW 10,000=USD 9.80) Dummy for respondent's household monthly electricity bill being less than or equal to KRW 30,000 (0=yes; 1=no)

0.50

0.50

13.91 4.16

2.42 2.00

0.73

0.44

statistic calculated under the null hypothesis is less than 0.01. In particular, the estimate for the spike is 0.615, which is similar to the proportion of the sample giving zero WTP responses provided in Table 1 (61.2%). This indicates that the spike model employed here fits our data well. Using Eq. (9) and the values presented in Table 2, we obtain an estimate of the mean monthly WTP of KRW 1402 (USD 1.37) per household. The t-value is 15.91 and thus the estimate is statistically significant at the 1% level. To address the uncertainty related to the computation of the estimate, we report the confidence intervals for the estimate. For this purpose, the parametric bootstrapping method suggested by Krinsky and Robb [65] is the method most widely employed in the literature. We use this method, with 5000 replications, to obtain the 95% and 99% confidence intervals, which are reported in Table 2. 4.3. Estimation results of the OOHB DC spike model with covariates We seek to estimate the spike model using the covariates explained in subsection 3.2. Some socioeconomic variables used for the covariates are defined in Table 3. These are related to the characteristics of the respondent reported in the third column of Table 3. The average years of schooling of respondents in this study was 13.91. The average monthly household income of the respondents in this study was KRW 4.16 million (USD 4.1 thousands). A total of four variables are contained in the model. The characteristics of the sample appear to reflect those of the population well as we drew a random sample of Korean households with the help of the professional polling firm. The results of estimating the spike model, including the variables shown in Table 3, are described in Table 4. The estimated coefficients for the Gender, Education, Income, and Bill variables are statistically significant at the 5% level. The estimated coefficient of Gender having a negative sign indicates that if respon-

Table 2 Estimation results of the spike model.

Constant Bid amounta Spike Mean WTP per household per month t-value 95% confidence intervalb 99% confidence intervalb Number of observations Log-likelihood Wald statistic (p-value)c

Mean

Bill

The estimation results of the OOHB DC spike model are reported in Table 2. The parameter estimates can be obtained by finding the parameter values maximizing Eq. (8), in other words, by applying the maximum likelihood estimation method. All the estimates for the two parameters, α and β , are statistically significant at the 1% level. Furthermore, the null hypothesis that the parameter estimates are all zero can be rejected at the 1% level in that the p-value for the Wald

Coefficient estimatesd

Definitions

Education Income

4.2. Estimation results for the OOHB DC spike model

Variables

Variables

Table 4 Estimation results of the spike model with covariates.

Notes: a Unit is KRW 1000 (USD 1 was approximately equal to KRW 1020 at the time of the survey). b Confidence intervals calculated using the Monte Carlo simulation technique of Krinsky and Robb [65] with 5000 replications. c Null hypothesis is that all the parameters are jointly zero and the corresponding pvalue is reported in parentheses beside the statistic. d Numbers in parentheses beside the coefficient estimates are t-values, computed from the analytic second derivatives of the log-likelihood. # Indicates statistical significance at the 1% level.

Variablesa

Coefficient estimates

t-values

Constant Bid amountb Gender Education Income Bill Wald statistic (p-value)c Log-likelihood Number of observations

−1.939 −0.353 −0.256 0.098 0.105 −0.300 609.10 (0.000) −983.77 1000

−4.71* −21.25* −1.97* 3.38* 3.36* −2.01*

Notes: a Variables are defined in Table 3. b Unit is KRW 1000 (USD 1 was approximately equal to KRW 1020 at the time of the survey). c Null hypothesis is that all the parameters are jointly zero and the corresponding pvalue is reported in parentheses beside the statistic. * Indicates statistical significance at the 5% level.

247

Renewable and Sustainable Energy Reviews 81 (2018) 242–249

S.-Y. Lim et al.

dents are female they will have a tendency to answer “yes” to a given bid. The lengthier a respondent's educational period, the higher the probability for responding “yes” to the WTP question. The finding that the coefficient estimate for Income has positive sign indicates that the respondents with higher incomes are more likely to record a “yes” response to a presented bid. It implies that a sound national economy can positively affect the expansion of FC. The respondents with lower bills have a tendency to report “yes” response to a given bid. This implies that those who pay electricity bills less than or equal to KRW 30,000 (USD 29.4) are more likely to state “yes” to the bid. We can thus estimate that users of considerably less electricity will accept paying something for the expansion of FC power generation.

results from our study can be added to the external benefits of expanding FC power generation from 565 GWh in 2013 to 13,449 GWh by 2027. The mean additional monthly WTP is estimated to be KRW 1407 (USD 1.37) per household. This figure corresponds to about 3% of monthly household electricity bills, which is KRW 47,979 (USD 47.04). This can be interpreted as a premium of FC generation over conventional fossil fuel generation. If the additional FC generation cost arising compared to conventional fossil fuel generation is less than the premium, FC generation can successfully be implemented. However, if not, further action is needed to ensure success in implementing the FC generation expansion. For example, this might involve assigning FC in renewable portfolio standard and restoring the feed-in-tariff, which was abolished and replaced with the renewable portfolio standard as of 2012. Moreover, various tax credits and other incentives can be made available by the government to encourage expansion. A sound incentive for FC providers to develop the technologies for producing FC more cheaply and efficiently and expand FC-producing facilities might be the provision of subsidies. Information on the premium obtained from our study can be employed in determining the levels of tax reductions, tax credits, or subsidies to foster FC power generation.

4.4. Discussion of the results To obtain an estimate of a representative household's WTP, one must use the mean WTP estimate obtained from investigating sample observations and information on population size. In the course of this estimation, the most important issue is whether or not the sample is representative of the population. As addressed above, the sampling was conducted by a professional polling company to secure the randomness of the sampling and its consistency with the characteristics of the population. Another important issue is the response rate in the CV survey. Our CV survey was implemented using in-person face-to-face interviewing and thus the response rate was almost 100%. Thus, it cannot be denied that the sample used in this study is representative of the population. We use the mean WTP estimate from the model with no covariates as the setting of the covariates may influence the mean WTP value if we use the mean WTP value from the model with covariates. According to Korea National Statistical Office [35], the number of households in Korea was 18,457,628 at the time of the survey (i.e., July 2014). Using this information, and expanding the value to the national population, gives us KRW 310.5 billion (USD 304.4 million) per year. As shown in Table 5, the corresponding 95% and 99% confidence intervals for the total economic value are KRW 276.2 billion (USD 270.8 million) to KRW 351.5 billion (USD 344.6 million) and KRW 266.7 billion (USD 261.5 million) to KRW 365.9 billion (USD 358.7 million), respectively. Overall, we can conclude that Korean households are ready to bear a share of the financial burden to expand FC power generation. The results of our estimations have various potential uses. They can be incorporated in computing the total economic benefits of consuming electricity generated from FC. By consuming electricity, an individual gains benefits that are made up of private and external benefits. FC power can provide much larger external benefits in terms of reducing CO2 emissions, and ensuring environmental protection and energy supply security compared to fossil fuels. Private benefits derive from electricity consumption and can easily be measured. In contrast, the evaluation of external benefits is problematic because the external benefits arise outside the market. Notwithstanding this difficulty, the

5. Conclusions Recently, Korea, the seventh largest CO2 emitter in the world, announced its 2030 mitigation target of 37% from the business-asusual level. One of the effective alternatives for achieving this target is to expand the use of new and renewable energy sources. FC can provide more stable, predictable baseload electricity than other renewable energy, such as photovoltaic and wind turbine power, both of which are difficult to manage because of sudden fluctuations. Therefore, the Korean government plans to expand FC power generation significantly by 2027. This study has measured the public WTP for implementing FC power generation expansion policy, employing data gathered from a CV survey administered to 1000 randomly chosen Korean households. To mitigate the response effect in eliciting WTP and increase statistical efficiency in analyzing the WTP data, we used an OOHB DC question format. In addition, we employed the spike model to model zero WTP responses. These approaches were well administered in that all the parameter estimates were statistically significant at the 1% level. The mean additional monthly WTP for the expansion was computed to be KRW 1407 (USD 1.37) per household. Expanding the value to the national population gave us KRW 310.5 billion (USD 304.4 million) per year. We can conclude that Korean households are ready to shoulder some of the financial burden of expanding FC power generation. This study analyzed the households’ WTP for conducting the FC power generation expansion policy from the standpoint of CO2 emissions reduction. Thus, the costs that arise in terms of society and electric power generation companies were not dealt with. For example, it may be more important to measure both social marginal cost of reducing CO2 emissions and power companies’ private marginal cost for expanding FC power generation. Analyzing the marginal costs can give us useful and quantitative information on the optimal level of CO2 emissions reduction and/or power company's investment on CO2 emissions. As a second stage of our study, an extension of the scope of the study to the analysis of the marginal costs will produce interesting results.

Table 5 Estimation results of total willingness to pay (WTP) for expanding fuel cell power generation in Korea.

Mean yearly WTP per household

Estimates

95% confidence intervals

99% confidence intervals

KRW 16,824

KRW 14,964– 19,044 (USD 14.67– 18.67) KRW 276.2–351.5 billion (USD 270.8–344.6 million)

KRW 14,448– 19,824 (USD 14.16– 19.44) KRW 266.7–365.9 billion (USD 261.5–358.7 million)

(USD 16.49) Total annual WTP

KRW 310.5 billion (USD 304.4 million)

Acknowledgments This work was supported by the Korea Institute of Energy Technology Evaluation and Planning (KETEP) and the Ministry of Trade, Industry & Energy (MOTIE) of the Republic of Korea (No. 20164030201060).

Note: At the time of the survey, there were 18,457,628 households in Korea [35].

248

Renewable and Sustainable Energy Reviews 81 (2018) 242–249

S.-Y. Lim et al.

emissions trading scheme in Korea. Energy Policy 2015;83:82–6. [33] Kwak SY, Yoo SH, Kim CS. Measuring the willingness to pay for tap water quality improvement: results of a contingent valuation survey in Pusan. Water 2013;5:1638–52. [34] Arrow K, Solow R, Portney PR, Leamer EE, Radner R, Schuman H. Report of the NOAA panel on contingent valuation. Federal register. 58; 1993. p. 4601–14. [35] Korea National Statistical Office; 18th, April, 2016. Available at: 〈http://kosis.kr〉. [36] Hanemann WM. Welfare evaluations in contingent valuation experiments with discrete responses. Am J Agr Econ 1984;66:332–41. [37] Hanemann WM, Loomis J, Kanninen BJ. Statistical efficiency of double-bounded dichotomous choice contingent valuation. Am J Agr Econ 1991;66:1255–63. [38] McFadden D. Contingent valuation and social choice. Am J Agr Econ 1994;76:689–708. [39] Bateman IJ, Langford IH, Jones AP, Kerr GN. Bound and path effects in double and triple bounded dichotomous choice contingent valuation. Resour Energy Econ 2001;23:191–213. [40] Carson RT, Groves T. Incentive and informational properties of preference questions. Environ Resour Econ 2007;37:181–210. [41] Cooper JC, Hanemann M, Singorello G. One and one-half bound dichotomous choice contingent valuation. Rev Econ Stat 2002;84:742–50. [42] Lim KM, Lim SY, Yoo SH. Estimating the economic value of residential electricity use in the Republic of Korea using contingent valuation. Energy 2014;64:601–6. [43] Kwak SY, Yoo SH. The public's value for developing ocean energy technology in the Republic of Korea: a contingent valuation study. Renew Sustain Energy Rev 2015;43:432–9. [44] Cameron TA, James MD. Efficient estimation methods for “closed-ended” contingent valuation surveys. Rev Econ Stat 1987;69:269–76. [45] McConnell KE. Models for referendum data: the structure of discrete choice models for contingent valuation. J Environ Econ Manag 1990;18:19–34. [46] Kriström B. Spike models in contingent valuation. Am J Agr Econ 1997;79:1013–23. [47] Yoo SH, Kwak SJ. Using a spike model to deal with zero response data from double bounded dichotomous choice contingent valuation surveys. Appl Econ Lett 2002;9:929–32. [48] Lee JS, Yoo SH. Willingness to pay for GMO labeling policies: the case of Korea. J Food Saf 2011;31:160–8. [49] Jang J, Lee J, Yoo SH. The public's willingness to pay for securing a reliable natural gas supply in Korea. Energy Policy 2014;69:3–13. [50] Lim HJ, Yoo SH. Train travel passengers' willingness to pay to offset their CO2 emissions in Korea. Renew Sustain Energy Rev 2014;32:526–31. [51] Smith VK, Osborne LL. Do contingent valuation estimates pass a “scope” test? A meta-analysis. J Environ Econ Manag 1996;31:287–301. [52] Chien YL, Huang CJ, Shaw DA. General model of starting point bias in doublebounded dichotomous contingent valuation surveys. J Environ Econ Manag 2005;50:362–77. [53] Buschena DE, Anderson TL, Leonard JL. Valuing non-marketed goods: the case of elk permit lotteries. J Environ Econ Manag 2001;41:33–43. [54] Parsons GR, Myers K. Fat tails and truncated bids in contingent valuation: an application to an endangered shorebird species. Ecol Econ 2016;129:210–9. [55] Whitehead JC. Plausible responsiveness to scope in contingent valuation. Ecol Econ 2016;128:17–22. [56] da Costa CA, Santos JL. Estimating the demand curve for sustainable use of pesticides from contingent-valuation data. Ecol Econ 2016;127:121–8. [57] Gelo D, Koch SF. Contingent valuation of community forestry programs in Ethiopia: controlling for preference anomalies in double-bounded CVM. Ecol Econ 2015;114:79–89. [58] Lo AY, Jim CY. Protest response and willingness to pay for culturally significant urban trees: implications for contingent valuation method. Ecol Econ 2015;114:58–66. [59] Tobias B. Keeping up appearances: motivations for socially desirable responding in contingent valuation interviews. Ecol Econ 2013;87:155–65. [60] Longo A, Hoyos D, Markandya A. Sequence effects in the valuation of multiple environmental programs using the contingent valuation method. Land Econ 2015;91:20–35. [61] Bateman IJ, Munro A, Poe GL. Decoy effects in choice experiments and contingent valuation: asymmetric dominance. Land Econ 2008;84:115–27. [62] Champ PA, Flores NE, Brown TC, Chivers J. Contingent valuation and incentives. Land Econ 2002;78:591–604. [63] Svedsäter H. Economic valuation of the environment: how citizens make sense of contingent valuation questions. Land Econ 2003;79:122–35. [64] Desvousges W, Mathews K, Train K. An adding-up test on contingent valuations of river and lake quality. Land Econ 2015;91:556–71. [65] Krinsky I, Robb AL. On approximating the statistical properties of elasticities. Rev Econ Stat 1986;68:715–9.

References [1] World Bank; 1st, December, 2016. Available at: 〈http://www.worldbank.org〉. [2] United States Energy Information Administration. International energy outlook 2013. Washington, DC: U.S. Department of Energy Forrestal Building. EI-40. 20585; 2013. [3] Korean Ministry of Trade, Industry, and Energy. The 6th basic plan for long-term electricity supply and demand (2013–2027), Gwacheon, Korea; 2013. [4] Korea Ministry of Government Legislation. The act on the promotion of the deployment, use and diffusion of new and renewable energy; 2014. [5] Korea Energy Agency. 2014 Annual report; 2015. [6] Mekhilef S, Saidur R, Safari A. Comparative study of different fuel cell technologies. Renew Sustain Energy Rev 2012;6:981–9. [7] Sharaf OZ, Orhan MF. An overview of fuel cell technology: fundamentals and applications. Renew Sustain Energy Rev 2014;32:810–53. [8] Xu Q, Zhang F, Xu L, Leung P, Yang C, Li H. The applications and prospect of fuel cells in medical field: a review. Renew Sustain Energy Rev 2017;67:574–80. [9] Rahman SNA, Masdar MS, Rosli MI, Majlan EH, Husaini T, Kamarudin SK, et al. Overview biohydrogen technologies and application in fuel cell technology. Renew Sustain Energy Rev 2016;66:137–62. [10] Jaccard M. Fossil fuels and clean, plentiful energy in the 21st century: the example of coal. EIB Pap 2007;12:80–104. [11] Kamsamron J, Sorapipatana C. Assessing CO2 abatement cost for Thailand's power generation. J Sustain Energy Environ 2014;5:21–6. [12] Romejko K, Nakano M. Portfolio analysis of alternative fuel vehicles considering technological advancement, energy security and policy. J Clean Prod 2017;142:39–49. [13] Wilberforce T, Alaswad A, Palumbo A, Dassisti M, Olabi AG. Advances in stationary and portable fuel cell applications. Int J Hydrog Energy 2016;41:16509–22. [14] Ju HC, Yoo SH. The environmental cost of overhead power transmission lines: the case of Korea. J Environ Plan Manag 2014;57:812–28. [15] United States Department of Energy. Comparison of fuel cell technologies. Energy efficient and fuel cell technologies program; February 2011. [16] Sevencan S, Lindbergh G, Lagergren C, Alvfors P. Economic feasibility study of a fuel cell-based combined cooling, heating and power system for a data centre. Energy Build 2016;111:218–23. [17] McLarty D, Brouwer J, Ainscough C. Economic analysis of fuel cell installations at commercial buildings including regional pricing and complementary technologies. Energy Build 2016;113:112–22. [18] Li H, Tian Y, Zuo W, Zhang J, Pan X, Li L, et al. Electricity generation from food wastes and characteristics of organic matters in microbial fuel cell. Bioresour Technol 2016;205:104–10. [19] Raza R, Akram N, Javed MS, Rafique A, Ullah K, Ali A, et al. Fuel cell technology for sustainable development in Pakistan – an over-view. Renew Sustain Energy Rev 2016;53:450–61. [20] Yoo SH, Kwak SJ, Kim TY. Assessing benefits from greenhouse gas emission reduction policy: a pilot case study of Korea. Int J Environ Pollut 2001;15:553–67. [21] Lee JS, Yoo SH, Kwak SJ. Public's willingness to pay for preventing climate change. Appl Econ Lett 2010;17:619–22. [22] Yoo SH, Kwak SJ. Willingness to pay for green electricity in Korea: a contingent valuation study. Energy Policy 2009;37:5408–16. [23] Lee CY, Heo HJ. Estimating willingness to pay for renewable energy in South Korea using the contingent valuation method. Energy Policy 2016;94:150–6. [24] Hanley N, Nevin C. Appraising renewable energy developments in remote communities: the case of the North Assynt Estate. Scotl Energy Policy 1999;27:527–47. [25] Zarnikau J. Consumer demand for ‘green power’ and energy efficiency. Energy Policy 2003;31:1661–72. [26] Nomura N, Akai M. Willingness to pay for green electricity in Japan as estimated through contingent valuation method. Appl Energy 2004;78:453–63. [27] Wiser RH. Using contingent valuation to explore willingness to pay for renewable energy: a comparison of collective and voluntary payment vehicles. Ecol Econ 2007;62:419–32. [28] Li H, Jenkins-Smith HC, Silva CL. Public support for reducing US reliance on fossil fuel: investigating household willingness to pay for energy research and development. Ecol Econ 2009;68:731–42. [29] Soliño M, Vázquez MX, Prada A. Social demand for electricity from forest biomass in Spain: does payment periodicity affect the willingness to pay?. Energy Policy 2009;37, [351–40]. [30] Heo JY, Yoo SH. The public's value of hydrogen fuel cell buses: a contingent valuation study. Int J Hydrog Energy 2013;38:4232–40. [31] Mitchell RC, Carson RT. Using surveys to value public goods: the contingent valuation method. Washington, DC: Resources for the Future; 1989. [32] Song TH, Lim KM, Yoo SH. Estimating the public's value of implementing the CO2

249