Environmental pricing of externalities from different sources of electricity generation in Chile

Environmental pricing of externalities from different sources of electricity generation in Chile

Energy Economics 34 (2012) 1214–1225 Contents lists available at SciVerse ScienceDirect Energy Economics journal homepage: www.elsevier.com/locate/e...

568KB Sizes 0 Downloads 28 Views

Energy Economics 34 (2012) 1214–1225

Contents lists available at SciVerse ScienceDirect

Energy Economics journal homepage: www.elsevier.com/locate/eneco

Environmental pricing of externalities from different sources of electricity generation in Chile Claudia Aravena a, d,⁎, W. George Hutchinson a, b, c, 1, Alberto Longo a, b, c, 1 a

Gibson Institute for Land, Food and the Environment, School of Biological Sciences, Queen's University Belfast, Medical Biological Centre, 97 Lisburn Road, Belfast BT9 7BL, UK UKCRC Centre of Excellence for Public Health (NI), Queen's University Belfast, UK Institute for a Sustainable World, Queen's University Belfast, UK d Department of Economics, Trinity College Dublin, Arts Building, Dublin 2, Ireland b c

a r t i c l e

i n f o

Article history: Received 24 September 2010 Received in revised form 21 August 2011 Accepted 6 November 2011 Available online 12 November 2011 JEL classification: C25 D12 Q42 Q51 Keywords: Contingent valuation Externalities of electricity generation Fossil fuels Large scale hydropower Willingness to pay for renewable energy

a b s t r a c t The rapid increase in electricity demand in Chile means a choice must be made between major investments in renewable or non-renewable sources for additional production. Current projects to develop large dams for hydropower in Chilean Patagonia impose an environmental price by damaging the natural environment. On the other hand, the increased use of fossil fuels entails an environmental price in terms of air pollution and greenhouse gas emissions contributing to climate change. This paper studies the debate on future electricity supply in Chile by investigating the preferences of households for a variety of different sources of electricity generation such as fossil fuels, large hydropower in Chilean Patagonia and other renewable energy sources. Using Double Bounded Dichotomous Choice Contingent Valuation, a novel advanced disclosure method and internal consistency test are used to elicit the willingness to pay for less environmentally damaging sources. Policy results suggest a strong preference for renewable energy sources with higher environmental prices imposed by consumers on electricity generated from fossil fuels than from large dams in Chilean Patagonia. Policy results further suggest the possibility of introducing incentives for renewable energy developments that would be supported by consumers through green tariffs or environmental premiums. Methodological findings suggest that advanced disclosure learning overcomes the problem of internal inconsistency in SB-DB estimates. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Pressures from economic and population growth, particularly in developing countries, have traditionally pushed governments to look at cost-effective options, mainly those exploiting fossil fuel sources, to cope with the increased demand for electricity from households and industry. Recent studies have shown that electricity generation from fossil fuels is one of the main sources of greenhouse gas emissions contributing to climate change (IPCC, 2007), and that associated environmental and social costs are rarely internalized in cost-benefit analyses and project evaluations (Färe et al., 2010; Georgakellos, 2010). Concerns over environmental issues and future energy security have boosted governments' interests in looking at the potential of

⁎ Corresponding author. Tel.: + 44 28 9097 2685; fax: + 44 28 9097 5877. E-mail addresses: [email protected], [email protected] (C. Aravena), [email protected] (W.G. Hutchinson), [email protected] (A. Longo). 1 Tel.: + 44 28 9097 2321. 0140-9883/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.eneco.2011.11.004

renewable energy sources (RES) 2 to meet the rising energy demand. Although RES are generally more expensive than traditional fossil fuels, they are recognized for entailing lower environmental and social impacts (European Commission, 2003, 2005). 3 This study focuses on the electricity sector in Chile and shows that consumers share these concerns in their willingness to pay (WTP) for electricity from more or less environmentally damaging sources. To cope with the rising demand for electricity in Chile, the three main options are: (i) Increase the use of fossil fuels, such as coal, oil

2 It is important to clarify at this stage the use of the term RES through the paper. In Chile, RES are classified into two categories: Conventional and Non-Conventional. The development of large scale hydropower projects, entailing the construction of large dams, is considered as conventional RES. Non-conventional RES are wind, solar, biomass, wave and geothermal power, as well as small-scale hydropower developments. In this paper we consider as “RES” the non-conventional sources including wind, solar and biomass power. 3 The research on the externalities of energy production (ExternE) carried out by the European Commission (2003, 2005) shows that external costs differ according to the sources used to produce energy, the technology, the location of the plants, the dispersion models used to assess the affected areas, and the physical characteristics of a country.

C. Aravena et al. / Energy Economics 34 (2012) 1214–1225

and gas, to feed new thermoelectric plants; (ii) Develop large-scale hydroelectric dams in Chilean Patagonia; or (iii) Introduce and boost the construction of a network of RES. In Chile, investments in RES have mainly been targeted to conventional projects, such as largescale hydropower. Non-conventional projects, such as wind, solar, biomass, wave and geothermal power plants are less common. However, large hydropower projects, despite their ability to provide large quantity of electricity with low greenhouse gas emissions, are often criticized for disrupting ecosystems, spoiling landscapes and causing other environmental and social impacts (Miller and Spoolman, 2009). Public concern and protests against large hydropower projects in Chile, have caused delays and an increase in costs, and have highlighted the importance of the debate about future sources and environmental pricing of electricity supply (Hall et al., 2009). As Chile has a great potential in natural energy resources (Hall et al., 2009), it is therefore essential to learn about Chilean households' preferences for developing alternative sources of electricity generation. This paper investigates the public's preferences and attitudes to different sources of electricity production in Chile and the willingness to pay (WTP) for RES. It considers the public's preferences with respect to environmental pricing and how externalities of electricity generation could be internalized and influence future energy policies. By using a Contingent Valuation survey (Bateman et al., 2002; Mitchell and Carson, 1989), we assess what premium households are willing to pay for RES compared to electricity, either from large hydropower dams in Chilean Patagonia or from new thermoelectric plants in the central region of Chile. The results can be used to inform policymakers on the choice of future sources of electricity and to design energy programs which take into account social and environmental goals. Moreover, the results consider not only environmental externalities but also externalities related to energy security and reliance on imported fuel stocks (Owen, 2006). Past research has investigated consumers' WTP premiums for renewable energy in industrialized countries such as the United States (Byrnes et al., 1999; Farhar, 1999; Farhar and Coburn, 1999; Farhar and Houston, 1996; Groothuis et al., 2008; Roe et al., 2001; Whitehead and Cherry, 2007; Wiser, 2007; Zarnikau, 2003); the United Kingdom (Batley et al., 2000, 2001; DiazRainey and Ashton, 2007; Hanley and Nevin, 1999), Italy (Bollino, 2009; Polinori, 2009), the Netherlands (Verbeet, 2007), Greece (Koundouri et al., 2009), Canada (Rowlands et al., 2003), Australia (Ivanova, 2005) and Japan (Nomura and Akai, 2004). All these studies have applied the contingent valuation method. Some studies have used choice experiments to consider the environmental benefits of RES and the environmental pricing of externalities (e.g. Bergmann et al., 2006; Borchers et al., 2007; Longo et al., 2008 consider a set of RES as a whole; Álvarez-Farizo and Hanley, 2002; Campbell et al., 2011; Ek, 2002; Meyerhoff et al., 2010, focus on wind power in Spain, Sweden, Germany and Chile respectively; and Sundqvist, 2002, on hydropower in Sweden). All these studies, despite their differences in design, find that consumers generally place a significant premium on RES and impose a value on the externalities generated by the different means of energy generation sources. Most of the countries concerned (Australia, Austria, Belgium, Canada, Germany, Ireland, Finland, Norway, Sweden, Switzerland, the Netherlands, the UK and the US), have already established green tariffs for RES. Nevertheless there are still many Latin American and developing countries with great potential for developing RES that remains unexplored. Furthermore, environmental and social laws are generally weaker in these countries and the assessments of energy projects do not usually consider the value of the externalities associated with energy generation. In addition, only a few previous studies (Bollino, 2009; Georgakellos, 2010; Ivanova, 2005; Polinori, 2009; Scarpa and Willis, 2010) 4 establish comparison among costs and 4

Georgakellos (2010) uses a different methodology based on external data instead of stated preferences to account for the externalities of fossil fuels. He concludes that electricity prices would increase with the consideration of these costs.

1215

premiums for RES. These studies, except Georgakellos (2010) find that environmental premiums are not enough to cover the additional costs of producing renewable energy. This research aims to contribute to this debate over whether electricity price premiums for RES over electricity from fossil fuels and large dam hydropower cover the additional costs of this method of production and make RES a more competitive alternative. It also considers whether environmental prices should be introduced in energy related cost-benefit analysis. While this paper aims primarily at policy analysis, it makes a methodological contribution in terms of a novel test for the effect of advanced disclosure Learning Design Contingent Valuation (LDCV) (Bateman et al., 2004, 2008; Cooper et al., 2002) — an approach where respondents are presented with repeated valuation tasks as a way to learn the double bounded dichotomous choice (DBDC) mechanism and gain experience of their underlying WTP. The LDCV is shown to produce internal consistency of single bounded and double bounded contingent valuation estimates of WTP produced from the same dataset. The possibility of internal inconsistency in WTP has been an enduring criticism of what is otherwise the preferred method of contingent valuation. The rest of the paper is organized as follows. In Section 2 characteristics of the Chilean electricity sector are presented and the future of the national electricity supply is discussed. Section 3 describes the methodology and survey design, followed by the presentation of the results in Section 4. Finally, Section 5 concludes the paper. 2. Background: the Chilean energy sector Electricity demand in Chile has increased by 6% per annum over the last decade, and it is expected to continue this level of increase in the coming years due to population growth and development of economic activities, especially in the industrial and mining sectors. The National Energy Committee of Chile (NEC) has estimated that on average the increase in the demand by consumers linked to the major electrical grid in the country (the Central Interconnected System — CIS) will be about 7% per annum for the next 25 years from 2007. 5 Currently, 58% of Chile's electricity is generated using thermoelectric sources (fossil fuels), specifically oil, coal, and gas. These sources generate CO2 emissions, contributing to climate change (Färe et al., 2010; Georgakellos, 2010). They also imply a high dependence on energy imports. The remaining 42% of electricity comes from hydropower, including small and large dams. There have been a number of serious conflicts over the construction of large dams in environmentally and culturally sensitive areas. The latest two projects (Pangue and Ralco Dams in the Bio Bio Region) encountered strong social oppositions and stricter environmental regulations, which resulted in construction delays and increased costs. At the moment almost no electricity is being generated in Central Chile by RES. At a national level, in 2008, RES generation accounted for 0.07% of the total energy supply of the country, mainly coming from wind power. 6 The current alternatives to meet increased energy needs in Chile are: (i) Large hydropower developments in Chilean Patagonia, (ii) Increased use of fossil fuels (oil, coal and gas) in Central Chile, and (iii) Introduction of RES (wind, biomass and solar power). Due to the impacts associated with current sources, the government has pointed out the importance of extending and diversifying the energy mix and promoting RES. The target is to reach 10% of national generation based on these sources by 2024 (Chilean General Law of Electric Services Number, 20257, 2008). The social and environmental impacts of these alternatives are distinct, as are the costs associated 5 NEC report on electricity prices (2007). Available at www.cne.cl and statistics of CDEC — Centro de Despacho Económico de Carga (Economic Center of Charge Service) http://www.cdec-sic.cl/. 6 Source: Operation Statistics (1999/2008). CDEC-SIC.

1216

C. Aravena et al. / Energy Economics 34 (2012) 1214–1225

with their installation. RES are recognized for being more environmentally friendly, but their costs are higher than their counterparts. Furthermore, current incentives and support for RES in Chile are low and there are no green schemes oriented towards encouraging customers to support green energy and encouraging the utilities to produce it. 7 This study focuses on customers of one of the four electrical grid systems that cover Chile, the Central Interconnected System (CIS). The CIS is the most important electrical grid in the country, with an installed capacity of 9000 MW, representing 71% of the total installed power in the country, in the area where the main cities and national industries are located. The area covered by the CIS corresponds to 43% of the total Chilean territory, and provides electricity to 93% of the total population of Chile. Endesa, the major electricity utility in the country has presented a project to build five hydroelectric plants with large dams and reservoirs on two rivers in the Aysén Region (Chilean Patagonia): the Baker and the Pascua, both of which have river basins with the richest biodiversity in the country. These hydropower projects would generate approximately 2750 MW, which would be incorporated into the CIS. This project would flood more than 5000 ha of lands, currently used for agriculture, recreation, tourism, and biodiversity conservation zones. In particular, some endangered species such as the huemul, a typical Chilean deer, would be affected. The rivers would be dammed, impacting water activities and some ecological processes and functions. In addition, the project aims to install a network of pylons crossing the Patagonian landscape, thus spoiling the scenic view to tourists visiting the area. All the electricity produced would be transferred to the CIS and no energy generation would be supplied to the Aysen Region. Opposition of ecological groups, inhabitants, NGOs and others to this project has recently arisen, and efforts to find alternatives have increased. A second major option to cope with the increased demand for electricity is to build new thermoelectric plants in the Central part of Chile to supply the needed electricity to the CIS. These projects are based mainly on the use of gas and coal, with few of them using oil. This option would further increase Chile's dependency on imports of fossil fuels and increase air pollution and green house gas emissions. Finally, the government is evaluating projects using RES such as wind farms, biomass, solar and some geothermal power in North and Central Chile. These projects have the potential to be more environmentally friendly, but their costs are higher. Most of the electricity generated by these projects will also be supplied to the CIS. 8

3. Survey design We use the contingent valuation method (CVM) (Bateman and Willis, 1999; Bateman et al., 2002; Hanemann, 1984; Mitchell and Carson, 1989) to elicit the willingness to pay (WTP) a premium for electricity supplied from a mix of RES, consisting of wind, solar and biomass power in Chile, all of which we assure respondents have lower environmental impacts than conventional options. Two scenarios were presented in the same questionnaire, one after the other. In both scenarios respondents were asked their WTP a premium for introducing and developing renewable energy instead of using an alternative conventional source. In the first scenario, the conventional alternative was electricity from hydropower produced by large dams in Chilean Patagonia. In the second scenario the alternative was

7 After the completion of this study in March 2008, the Chilean government issued the plan to encourage electrical utilities to generate 10% of their energy for 2024 by using RES (Chilean General Law of Electric Services Number, 20257, 2008). 8 Although nuclear power is a potential alternative for Chile, the government has committed to not introduce it for the foreseeable future. For this reason, this source has been excluded from this study.

energy from fossil fuels based on oil, coal and gas. All respondents faced the same choices with the alternatives presented in that order. 9 The scenario presented to respondents included information about the Chilean energy market, the rising electricity demand and the need to increase electricity supply. After that, it described the hydroelectric projects in Chilean Patagonia with their environmental and social impacts. The impacts presented in this scenario were: landscape intrusion; flooding of pristine territories which host a large variety of flora and fauna; installation of a network of pylons crossing natural reserves; impact on sports that currently use the river (e.g. kayaking, rafting and fishing); negative impacts on agriculture, tourism and fishing and relocation of displaced inhabitants. Positive impacts, including irrigation and establishment of new tourist resources such as artificial reservoirs created by dams were also explained. Besides the description of the hydropower projects and their impacts, the projects for electricity generation by RES were presented explaining also their associated benefits and impacts. The impacts considered for RES were: landscape intrusion and air pollution caused by some biomass generation plants. After the presentation of this scenario, respondents were asked how they would vote if an alternative project for additional electricity generation based on RES were offered at a determined extra cost instead of large dams in Chilean Patagonia. This extra cost would correspond to the WTP premium for RES. This part was followed by the presentation of the second scenario, which considered the description of the thermoelectric projects in the central part of Chile and the alternative of RES. Once again, the environmental and social impacts related to these sources were explained. In the case of thermoelectric projects, the impacts considered were: greenhouse gas emissions contributing to global warming; health problems; landscape intrusion; and dependence on imported energy suppliers. This was followed by questions eliciting the WTP for introduction and development of RES instead of fossil fuels. The respondents were asked how they would vote if they were offered the RES project at extra cost as an alternative to additional electricity from fossil fuels as their baseline. Respondents were asked to treat each scenario independently, i.e. as if this were the only scenario which could be implemented; thus the second valuation question would not add another extra cost to the first one presented. The interviewers used graphics illustrating the different scenarios, as well as pictures of the current electricity generation plants and plants for RES.10 While eliciting WTP for a single outcome – RES – against two alternative baselines is an innovative use of contingent valuation, it is entirely realistic in this situation where we are considering future electricity supply and the future baselines production methods are not yet determined. The payment vehicle was, similar to that noted in many related studies (Borchers et al., 2007; Byrnes et al., 1999; Ivanova, 2005; Zarnikau, 2003; among others), an additional cost on the fixed part of the monthly electricity bill. In order to minimize the effect of hypothetical bias we followed the recommendations of Cummings and Taylor (1999) on “Cheap Talk” design for the contingent valuation method. The wording used in the cheap talk is the following: “Experiences from similar surveys show that often people answer in one way but actually act differently. Sometime people say they are willing to pay a different amount of money than they would actually pay, maybe because they do not think of the real impact 9 The order of the two valuation questions was not switched due to sample size constrains. Switching them would imply the need to split the sample into two subsamples, and the objective of the study does not consider the test for order effects. To minimize any ordering effects, respondents were given information or advance disclosure of the valuation exercises before they carried out the contingent valuation tasks. Bateman et al. (2004) have shown that the introduction of advance disclosure of the sequence of goods to be valued (also called visible choice set) eliminates effectively the order effect. 10 The pictures were carefully chosen and tested in order to avoid the introduction of bias.

C. Aravena et al. / Energy Economics 34 (2012) 1214–1225

it has on the family budget. We would like you to think seriously on this point and answer as if you would really have to pay the amount of money you will be asked for, considering that it would reduce your money available for buying other goods”. This text has been shown by Cummings and Taylor (1999) to eliminate hypothetical bias effectively. Similar findings have been shown by Whitehead and Cherry (2007), Aadland and Caplan (2006) and Carlsson and Martinsson (2006). The questionnaire was designed with information obtained from focus groups and several pilot studies conducted from September to December 2007. Furthermore, meetings with different professionals related to energy, ecology, economics, sociology and psychology were held. On the basis of the information collected in the focus groups and pilot studies a bid vector was designed following Scarpa and Bateman (2000) and Hutchinson et al. (2001). The bid vector extended from 200 CLP to 10,000 CLP and consisted of six levels, of which four were used as starting points.11 The initial bids were randomly allocated for each scenario valued. The final questionnaire was divided into four sections, preceded by the interviewer's introduction, which presented the research and the reason for the visit. The first section contained a number of ‘warm-up’ questions to assess respondents' opinion, perception, attitude and knowledge about different kinds of electricity sources, as well as questions about whether respondents have ever visited Chilean Patagonia. The second section included the presentation of scenarios and the contingent valuation (CV) exercises as described previously. The third section contained background questions on the household. These questions included the number of people that constitute the household, their gender, the number of children, the level of education of the respondent and household income, among others. After completing the interview, in the final section of the questionnaire the interviewers would note their own impressions about respondents' understanding of the valuation exercise, interest in the topic and other external factors that might have affected the interview. The valuation questions that followed each scenario were presented in a double-bounded dichotomous choice (DBDC) referendum format. This differs from the widely used single bounded format in that it adds a further follow-up question to the first referendum. The DBDC was initially developed by Carson et al. (1986) and consists of asking subjects to vote yes or no to implement the RES project at an initial bid price as if they were voting in a referendum. If the individuals votes yes they are asked to vote in a follow-up referendum on a higher bid price for the project. If they vote no, the follow-up referendum offers a lower bid price (Carson et al., 1986; Hanemann et al., 1991). An example of the bid schemes used in our study is shown in Fig. 1. The DBDC mechanism allows us to collect more information from each respondent (Hutchinson et al., 2001) and in consequence improves the statistical efficiency of the parameters compared with the single bounded mechanism. Although the single bounded elicitation format has been recommended by Arrow et al. (1993) and Carson et al. (2003) and described as an incentive compatible mechanism (Carson and Groves, 2007); the double bounded elicitation format, has gained popularity because it is shown to improve the statistical efficiency of estimates in CVM (Alberini, 1995; Hanemann et al., 1991; Kanninen, 1993). However, a well known problem with the double bounded referendum method, which was raised even in the initial literature (Hanemann et al., 1991) is the considerably higher WTP estimates produced by a single bounded (SB) analysis of the first bid compared to the double bounded (DB) analysis applied to the first and follow-up bids. This difference in WTP is referred to as “stylized fact” in an influential review paper (Carson and Groves, 2007) and represents one of the main criticisms of this elicitation format (DeShazo, 2002). This internal inconsistency between the

11

1 USD = 470 Chilean Pesos (CLP) at the time of the survey.

1217

two estimates has been widely recognized and documented in several studies (Cameron and Quiggin, 1994; Kanninen, 1995; McFadden, 1994). Cameron and Quiggin (1994) introduce a bivariate probit analysis including a parameter that captures the correlation between the answers to the first and second referendum, finding that there are not perfect correlations between the two responses. McFadden (1994) finds parametric and non-parametric evidence of internal inconsistency between responses to the first and second bid in a DBDC referendum elicitation format applied to preservation of wilderness areas in the US. Kanninen (1995) studies the biases in SBDC and DBDC related to bid design, sample size and the presence of outliers. Alberini et al. (1997) introduce a model specification that considers structural adaptations to the DBDC that are consistent with response incentives. Bateman et al. (2001) study the bound and path effects in multiple bounded dichotomous choice contingent valuation, finding strong evidence on the internally inconsistency of responses derived from multiple dichotomous choice questions. Finally, Cooper et al. (2002) proposed the one-and-one-half-bound dichotomous choice method in which they provide respondents with advance information related to the valuation institution and the range in which the cost of the good lies. In this way they attenuate the surprise element that potentially creates the SB-DB discrepancy. Similarly, papers using economic experiments have explored behavioral explanations for this anomaly (Burton et al., 2003, 2009) and an innovative field study using a sequence of valuations termed Learning Design CV (LDCV) has been shown to significantly attenuate this problem (Bateman et al., 2008). The LDCV consists of presenting repeated valuation tasks to respondents who would learn the way the DBDC mechanism works and gain experience of their underlying WTP, reducing the gap between the two responses. This latter paper also introduced a straight forward Monte Carlo test for significant differences between SB and DB estimates, which takes account of covariance, and demonstrates that a Sequential Learning Design can result in SB-DB differences which are not statistically significant. Following the ideas of Bateman et al. (2004, 2008) and Cooper et al. (2002) we introduce in this paper an innovative learning process with advance disclosure and repetition and we show how this SB-DB anomaly can be eliminated cost-effectively. Our paper makes a further contribution to this approach by taking the idea of Advanced Disclosure Learning applied in Bateman et al. (2004) to scope sensitivity and applies it to the SB-DB difference in mean WTP. The original idea of Advanced Disclosure introduced in this paper consists of informing respondents, before they perform the valuation tasks, about the number of goods they are going to value and the sequence in which they will be presented. In our paper it is explained at the outset that the respondent will be asked to vote first on their WTP for RES over large dams in Chilean Patagonia and secondly on their WTP for RES over thermoelectric generation in the Central Chile. This process of Advanced Disclosure is shown to eliminate ordering effects in Bateman et al. (2004) where a sequence of goods is valued in various orders. 12,13 In addition, we explained the double bounded referendum mechanism in advance to all respondents as follows: “Because the exact cost of the projects is not known today, we will ask you to vote on 2 different costs for each project. These costs

12 The fact that the order in which a sequence of contingent valuation questions is presented may affect the estimated WTP values was already pointed out by Cummings et al. (1986, page 33), and also by Mitchell and Carson (1989, page 237). 13 Respondents were told that they will face two valuations. Firstly they will value RES against the hydropower alternative and secondly RES against fossil fuels. The wording used in this part of the questionnaire was: “In producing the 15% extra electricity required in Chile we are looking at how much more you would be willing to pay for this renewables alternative over the two other types of energy. In what follows you will be asked how you would vote if a referendum was held to choose between renewable energy and each of the other alternatives in this order: First, renewable energy versus hydropower and second renewable energy versus thermoelectric”.

1218

C. Aravena et al. / Energy Economics 34 (2012) 1214–1225

YES

YY

NO

YN

YES

NY

NO

NN

VOTE 2: WTP

YES

1000 CLP? VOTE 1: WTP 500 CLP? VOTE 2: WTP

NO

250 CLP?

Fig. 1. Example of double bounded dichotomous choice bid price scheme.

represent the range into which the actual cost should fall. In what follows, you will vote for or against each alternative. You are asked how you would vote if the good could be provided at one of the two cost. This is followed directly by a second vote on how you would vote if the good could be provided at the second of the two costs.” In this paper we test for the first time whether using Advanced Disclosure cLearning to value the first good will significantly attenuate the WTP difference between SB and DB estimates or whether these differences will only become non-significant for valuations of the second good which will benefit from the effects of both Advanced Disclosure and Sequential Learning. In both cases we will use the Monte Carlo type difference tests introduced in Bateman et al. (2008). The survey was conducted as a personal interview of households in the metropolitan area of the two largest cities of Chile, Santiago and Concepción. Both cities are important parts of the CIS grid. The person interviewed was the one responsible for paying the electricity bill. Households were selected from a random selection of streets from all areas of the city following a two stage random sampling procedure, stratified by socio-economic status.

4. Econometric model The CV responses are analyzed using the Discrete Choice Random Utility Model (Hanemann, 1984; McFadden, 1974), which assumes that respondents who face two choices or alternatives will choose the option that grants them the highest level of utility. Consider q as the environmental quality defining the alternatives. For a closedended CV model consisting of a binary choice between two alternatives or states, q = 0,1; in the original state (status-quo), where q = 0, the utility function for respondent j is U0j = U0j(0, yj, sj, ε0j), where y is the income, s the characteristics of the household and ε is the stochastic component. The utility function is composed by two components; a deterministic part which is observable to the researcher denoted by Uj = Uj(q, yj, sj) and a stochastic component, εj, making Uj a random variable. When a change in the environmental quality has occurred, the utility function of the final state, q = 1, is equal to U1j = U1j(1, yj − Bid, sj, ε1j), where Bid is the proposed payment offered to the respondent. These functions can equivalently be written as U0j = U0j(yj, sj, ε0j) and U1j = U1j(yj − Bid, sj, ε1j) (Hanemann, 1984). 14

14 An anonymous referee suggested the following alternative specification of the utility function: U = U(q, y|s), which allows environmental quality to differ across socioeconomic groups. However, in this paper we use the specification of Hanemann (1984) and Haab and McConnell (2002) specified in Eq. (1) as commonly used in contingent valuation.

The respondent will answer “yes” to the offered bid if:     U 1j yj −Bid; sj ; ε1j > U 0j yj ; sj ; ε0j :

ð1Þ

Due to the random part of utility not being observable, the model relies on probabilities. The probability that an individual answers yes to the offered Bid (Haab and McConnell, 2002) is: j    k PrðyesÞ ¼ Pr U 1j yj −Bid; sj ; ε1j > U 0j yj ; sj ; ε0j :

ð2Þ

For simplicity we assume a linear utility function. The deterministic part of the utility function is defined as V qj ¼ α q sj þ βq yj ; where m P α qk sjk , where k corresponds to the covariates introduced in α q sj ¼ k¼1

the estimations. Thus, Eq. (2) can be expressed as:   j   k PrðyesÞ ¼ Pr V 1j sj ; yj −Bid þ ε1j −V 0j sj ; yj −ε0j > 0 :

ð3Þ

Assuming the marginal utility of income in the two CV states is constant, and assuming also that α1 − α0 = α, and that εj = ε1j − ε0j the difference in utility between the two states becomes: V 1j −V 0j ¼ αsj −βBidj :

ð4Þ

Then, the probability that individual j answers yes to the proposed bid is:       Pr yesj ¼ Pr αsj −β Bidj þ εj > 0      ¼ Pr − αsj −β Bidj b εj      > εj ¼ 1−Pr − αsj −β Bidj   ¼ Pr εj b αsj −βðBidÞ   ¼ F β′ x

ð5Þ

where F corresponds to the cumulated density function and x corresponds to the bid vector and a set of covariates to be included in the estimation. The random component of the utility function is assumed to be identically and independently (IID) distributed with a mean equal to zero. In order to obtain greater statistical efficiency in estimation, we estimate a Double Bounded Dichotomous Choice (DBDC) model (Alberini, 1995; Haab and McConnell, 2002). For the DBDC format four responses are possible: (yes, yes), (yes, no), (no, yes) and (no, no) as presented in Fig. 1. Therefore,

C. Aravena et al. / Energy Economics 34 (2012) 1214–1225

the probability associated with each of the responses in the individual choice is:   Pr ðyes; yesÞ ¼ 1−F β′ xH     ′ ′ Pr ðyes; noÞ ¼ F β xH −F β x     Pr ðno; yesÞ ¼ F β′ x −F β′ xL   Pr ðno; noÞ ¼ F β′ xL

ð6Þ

where the subscript H corresponds to the higher bid offered and L corresponds to the lower bid offered. From this specification, the log likelihood function of the model is given by: LL ¼

Nyy X

yy logðprobðyyÞÞ þ

i¼1 N ny

þ

X

N yn X

ny logð probðnyÞÞ þ

i¼1

nn logð probðnnÞÞ

design where values are obtained for mean WTPm and mean WTPn from the same sample of respondents. This method was originally outlined in Bateman et al. (2008) and involves repeated sampling with replacement from the original sample of households (the primary sampling unit) thus implicitly taking account of the co-variance between these samples. This technique also controls for nonindependence of values from models using the same sample of households and allows for non-normal distributions which are common in non-market valuations. If Δm,n is the estimate of difference in WTP, then for a single bootstrap sample b drawn with replacements, the estimate from the bth sample is Δm,n (b). The estimated standard error of Δm,n uses b bootstrap samples each of sample size equal to the original sample size N. This is given by: 1=2     2   : =ðb−1Þ se Δm;n ¼ ∑ Δm;n ðbÞ−Δ m;n

yn logð probðynÞÞ

i¼1 Nnn X

1219

ð12Þ

ð7Þ

i¼1

where yy, yn, ny, nn represent the responses of the individual and Nyy, Nyn, Nny, Nnn correspond to the number of occurrences of each of the response. We use a logistic cumulated density function for F and consequently Double Bounded Dichotomous Logit Models are estimated. The models in this paper are estimated with covariates to study the determinants of WTP. 5. Hypotheses tests for policy purposes In order to test for any differences in WTP for RES over the other two scenarios presented, we use a pair wise bootstrap method, to obtain the distribution of differences between welfare measures following Bateman et al. (2008). 15 This method is applied instead of a standard t-test to control for the non-independence in the values obtained for the welfare measures. The reason for this is because both valuations were presented to all respondents in the same survey task, generating a non-independent sample. 16 The aim of this method is firstly to test if there is a significant premium for RES over other sources of generation and secondly to test if the premiums in mean WTP for RES in the two scenarios are significantly different. The null hypothesis in the first case is:

In this study an empirical distribution of differences Δm,n is created using 10,000 replications of E[WTPm] − E[WTPn]. We use the percentile method to calculate confidence intervals and analyze the differences in the mean WTP. An empirical confidence interval at the 5% significance level is composed by removing 2.5% of the probability from each tail of the Δm,n distribution. By inspecting the results of the confidence intervals and the distributions of differences Δm,n represented by histograms, it is possible to accept the null hypothesis if the 95% confidence interval for Δm,n is composed of strictly positive or strictly negative values, meaning that the difference in the RES premiums over the other two methods of generation Δm,n is strictly positive or negative or to reject the null hypothesis on the basis that the difference in the RES premiums Δm,n is composed of both negative and positive values. 6. Hypotheses tests on advanced disclosure learning and internal consistency of SB-DB values

Ho : E½WTP m  ¼ 0

ð8Þ

As a validation of the DBDC methodology we test at the outset for the effects of Advanced Disclosure Learning on the internal consistency of the SB and DB estimates. We do this for the first good valued (m) which is WTP for RES over large hydropower. We use a similar pair wise bootstrap of differences method to test for statistically significant differences between SB and DB estimates as follows:

Ho : E½WTP n  ¼ 0

ð9Þ

E½WTP mSB –E½WTP mDB ≠0:

where m represents the first scenario (RES versus hydropower in Eq. (8)) and n represents the second scenario (RES versus fossil fuels in Eq. (9)). The null hypothesis of significant difference between RES premiums E[WTPm] and E[WTPn] then is: Ho : E½WTP m ≠E½WTP n :

ð10Þ

This hypothesis can be rewritten as: Δm;n ¼ E½WTP m –E½WTP n ≠0:

ð11Þ

The non-parametric bootstrap (Efron and Tibshirani, 1993) is used to obtain the distribution of differences Δm,n, hence computing the standard error of differences se(Δm,n) controlling for the sampling 15 An alternative method for testing differences in a maximum of two nonindependent valuations is given in Poe et al. (1997). 16 In calculating the variance of the differences given by var(WTPm − WTPn ) = var(WTPm) + var(WTPn) − 2 ∗ cov(WTPm − WTPn), the problem arises in that the cov(WTPm − WTPn) is not zero, and consequently the use of a statistical test for independent samples would provide incorrect results.

ð13Þ

The findings of this test will indicate whether Advanced Disclosure of the mechanism and the goods to be valued can attenuate the SB-DB difference as previously demonstrated for LDCV (Bateman et al., 2008). In testing for significant differences between SB and DB estimates for the second good valued (n) we are testing whether it requires the addition of Sequential Valuation Learning to Advanced Disclosure to remove the statistical significance of the SB-DB difference as follows: E½WTP nSB –E½WTP nDB ≠0:

ð14Þ

If the test in Eq. (13) above is rejected, this will show that our new Advanced Disclosure Method can very cost effectively demonstrate internal consistency in our DBDC data sets which will add a novel validity test for this frequently used though sometimes criticized Stated Preference method. If the test in Eq. (14) is rejected while that in Eq. (13) is not rejected then we further demonstrate the finding of Bateman et al. (2008) that LDCV based on repetitive or sequential learning is necessary to attenuate this SB-DB value difference.

1220

C. Aravena et al. / Energy Economics 34 (2012) 1214–1225

Table 1 Descriptive statistics for socio-economic variables.

Table 2 Results of factual and attitudinal questions on energy sources.

Respondents characteristics Characteristics of the person in charge of paying the electricity bill in the sample. Average figures. Female Age Years of education (level of education) Household characteristics Average figures Number of members in household Number of children Monthly income (Chilean pesos) Electricity bill (Chilean pesos) Member of an environmental organization

48.4% 43 13 years

4 1 720,664 20,669 1.2%

7. Results 7.1. Descriptive results A face-to-face survey was conducted during January and February 2008. In total 726 responses were collected, but due to item non-response, 711 responses were available for analysis. Table 1 shows the descriptive statistics of the households and respondents of the sample in our survey. No significant differences were found in household characteristics when compared with population statistics from the Chilean National Survey (MIDEPLAN, 2006). In general terms, the sample represents quite closely the population under study. Table 2 shows the results for some factual and attitudinal questions on energy sources presented in the first section of the questionnaire. Most of the respondents (75%) could name at least one of the current sources of electricity generation in the country (70% of them mentioned hydropower, 11% thermoelectric and 2% other renewable energy sources). 58% of the sample have heard about the hydropower projects in Chilean Patagonia, mainly from television and newspapers, and many of the respondents (54%) have personally seen a dam. It is notable that only about 10% of the households have visited the Aysén Region, generally for tourism, but a higher proportion of them (21%) plan to visit it in the following year. Table 2 also presents household preferences for future development of energy sources in Chile. 17 Solar and wind power appear to be the most popular choices among households, receiving 61% and 57% support respectively. This result is consistent with previous research (e.g. Borchers et al., 2007; Diaz-Rainey and Ashton, 2007; Ivanova, 2005; Wiser, 2007) which found that solar power is the energy source with most support in different countries. The least preferred energy sources are fossil fuels (oil, coal and gas) and biomass. In the case of biomass, the low percentage of support might be explained by the lack of knowledge of respondents about this source. At the initial stage of the survey most respondents knew little about this source. Nuclear power also showed a low rate of acceptance (12%). In general, respondents do not support this source due to the risks associated with it. This result provides some insights regarding the existence of preferences for RES. A surprisingly high proportion (39%) supports the development of hydropower, even though the question did not specify between small hydropower, or large dams. Finally, 98% of the respondents think that Chile needs to start generating more electricity with its own resources in order to be independent from external energy providers.

17 Respondents were asked in the first part of the questionnaire which energy sources they think the government should support and develop in Chile for meeting the increase in the energy demand. No information or details about these sources were given to respondents before presenting this question.

Factual questions

% of those interviewed

Can name any current method of electricity generation Have seen a dam personally Have heard about the Patagonian hydro power projects Have visited the Aysen Region

75% 54% 58% 9.8%

Reason for visiting Aysen Tourism Work/Business Family Other Plans to visit Aysen Region

69% 14% 9% 11% 21%

Preferred sources for electricity generation Oil Gas Coal Hydropower Wind power Solar power Biomass Wave power Geothermal power Nuclear power

5% 12% 4% 39% 57% 61% 8% 20% 20% 11%

7.2. Policy results In total 663 responses were available for analysis after protest zeros were eliminated. 18 Protest answers were recognized through the inclusion of probes after the two valuation questions of each scenario for cases where respondents answered NN (i.e. ‘no’ to both valuation questions in each task). The principal reasons given for protests were that government and/or enterprises should pay for the development of RES projects. Table 3 reports the results of the estimation of the Double Bounded Dichotomous Choice logit model with covariates. 19 In the models of Table 3, “female” is a dummy variable setting the gender of the respondent. The variable has a value of one if the respondent is a woman and zero if the respondent is a man. The variable “age” represents the respondent's age in years. “Education” represents the years of education of the respondent related to the level of studies. “Income” corresponds to the mid-point of the total household's monthly income after taxes, considering bonus, pensions and any external salary. “Dam” is a dummy variable having the value of one if the individual has ever seen a dam and zero otherwise. “Heard about project” is also a dummy variable with the value of one if the person has heard about the hydroelectric projects in Chilean Patagonia and zero otherwise. Variables “plan visit” or “visited” refers to whether the person has plans to visit Chilean Patagonia in the following year or if she/ he has visited it respectively. These variables take on values of one if the person answers yes and zero otherwise. Finally, the variable “know source” corresponds to a dummy variable that takes on a value of one if the person recognizes any of the current sources of electricity generation. Other variables, such as number of household members, number of children and cost of the electricity bill, were 18 It is standard practice in the analysis of CV data studies to disregard from the analysis protest zero answers — respondents that vote against the proposed programs because they either reject some element of the hypothetical scenario, or the payment vehicle, but may actually have a positive WTP (Mitchell and Carson, 1989). Initially, 686 responses were available for the first scenario (RES instead of hydropower in Chilean Patagonia) and 664 responses for the second scenario (RES instead of fossil fuels). However, due to the fact that the number of observations must be the same in both samples in order to perform the pair wise bootstrap for comparing the mean willingness to pay in both scenarios, all the responses presenting at least one protest answer were eliminated from the sample under analysis. 19 Table A.1 in Appendix A of this paper presents the information on the frequency of responses to each bid vector for the two valuations.

C. Aravena et al. / Energy Economics 34 (2012) 1214–1225

1221

Table 3 Results single and double bounded dichotomous choice logit models with covariates (t-values in parentheses). RES instead of hydropower

RES instead of fossil fuels

Single bounded

Double bounded

Single bounded

Variable

Coefficient

Coefficient

Coefficient

Coefficient

Constant Bid Female Age Education Income Dam Heard about project Plan visit Visited Know source Mean WTP WTP St. error Log-likelihood function Sample size

1.67 (2.24) − 0.79 (− 7.19) 0.23 (1.09) − 0.02 (− 2.89) 0.07 (1.70) 0.0009 (3.19) 0.12 (0.53) − 0.004 (− 0.20) 0.57 (1.97) − 0.58 (− 1.58) 0.38 (1.56) 3560 (3038–4082) 266.21 − 291.37 663

1.44 (2.45) − 0.78 (− 18.15) 0.17 (1.02) − 0.02 (− 3.32) 0.08 (2.21) 0.0009 (3.90) 0.16 (0.86) − 0.04 (− 0.20) 0.54 (2.16) − 0.65 (− 2.16) 0.37 (1.88) 3401 (3156–3646) 106.94 − 704.76 663

0.09 (0.14) − 0.53 (− 10.62) − 0.28 (− 1.51) − 0.003 (− 0.45) 0.12 (3.28) 0.0003 (1.49) – – – – 0.63 (2.99) 4042 (3684–4400) 182.74 − 366.61 663

− 0.17 (− 0.32) − 0.46 (− 18.47) − 0.09 (− 0.59) − 0.005 (− 0.84) 0.11 (3.55) 0.0005 (2.57) – – – – 0.51 (2.90) 4014 (3672–4356) 174.58 − 690.28 663

1000 500

Frequency

1500

2000

and WTP. In this study, the gender variable (female) has no significant effect on the price premium for RES. There are mixed results in previous studies related to this variable; while Diaz-Rainey and Ashton (2007) show the same finding of no significant difference, Wiser (2007) and Zarnikau (2003) found a significant effect, with women and men willing to pay higher amounts for RES respectively. In our study, the age variable is only significant when the base alternative is the hydropower development in Chilean Patagonia. The negative sign of the parameter of age indicates that younger people are willing to pay more than older people. Similar findings are reported in Borchers et al. (2007), Byrnes et al. (1999), Diaz-Rainey and Ashton (2007), Wiser (2007) and Zarnikau (2003); where WTP for RES, is generally higher for younger people, but these results are related to the baseline in the respective country where the survey was administrated, which in most of cases was fossil fuels. In our case, we consider that younger people are more likely to visit Chilean Patagonia than older people are; thus, they would be willing to pay more in order to keep it in its pristine state. In the first valuation scenario, when RES are presented as an alternative to large dams in Chilean Patagonia, if households have previously seen a dam personally, or if they have heard about the dam projects, this had no significant effect on the

0

included in previous estimations, but were not found to be significant. The membership of environmental organizations was also not included, as the numbers of households with members participating in these organizations were too few to be significant. The mean WTP for the sample is obtained from the estimated parameters and the confidence intervals are calculated using the Delta Method. The models presented were selected according to the policy relevance of the variables included and the improvements in the fit of the model, using a log-likelihood ratio test. 20 As expected, the bid coefficient is negative and statistically significant at the 5% significance level in both models. Households on average are willing to pay 3401 CLP ($7.24 USD) extra in their monthly electricity bill to support the development of RES as an alternative to the construction of large dams in Chilean Patagonia, thus avoiding the social and environmental impacts related to this kind of project in that area. On the other hand, households are willing to pay on average 4014 CLP ($8.54 USD) for RES when the alternative is to increase electricity generation by using fossil fuels. These environmental premiums or mean WTPs for RES correspond to a realistic 16% and 19% of the average monthly electricity bill when large dams and fossil fuels are the baselines respectively. The environmental price premium for RES over fossil fuels or thermoelectric sources is higher than the premium over large hydropower. A pair-wise bootstrap of differences in the price premiums between the two valuation scenarios reveals that this difference is statistically significant at the 95% confidence level. Fig. 2 presents the distribution of the results of the pair-wise differences between the two scenarios showing that almost all differences in the distribution are greater than zero. These results indicate that the externalities associated with fossil fuels are perceived as more environmentally damaging than those arising from the construction of large scale hydropower. Households attach higher environmental values to the impacts from thermoelectric sources than from those associated with hydropower in Chilean Patagonia. In both models the WTP for environmental price premiums increases with higher levels of education, income and the knowledge of the current source of electricity generation. These findings are in line with previous research; for example, Wiser (2007) and Zarnikau (2003) find that the WTP for RES increases with income and higher education. Byrnes et al. (1999) also found that highly educated people tend to be willing to pay a larger premium for RES, and Diaz-Rainey and Ashton (2007) found a positive relationship between income

Double bounded

0

500

1000

1500

WTP Differences in Chilean Pesos (CLP) E(WTP)Hydropower –E(WTP)Thermoelectric:

a

613 CLP (248.01 to 1019.69)

Ho: Not rejected

a

CLP = Chilean Pesos

20

In addition to the socio-economic information, we only used variables related to visiting Patagonia and knowledge of dams and proposed dams projects only in the specification where we assess the WTP for RES over large dams in Chilean Patagonia.

Fig. 2. Pair-wise bootstrap of differences in WTP for RES over hydropower and for RES over fossil fuels. Ho: WTPHydropower − WTPThermoelectric ≠ 0 (95% confidence intervals in parenthesis).

1222

C. Aravena et al. / Energy Economics 34 (2012) 1214–1225

Table 4 Present value of WTP premiums and investment costs for energy scenario I to IV. Energy source

Scenario I

Scenario II

Scenario III

Scenario IV

Wind Biomass Solar Geothermal

90% 9% 1% 0%

75% 15% 10% 0%

50% 25% 25% 0%

80% 9% 1% 10%

Cost-benefit analysis RES versus large dams ($US millions) Cost of the RES Scenario Cost of large dams scenario (A) Difference between RES and large dams investment costs (B) WTP premium for RES instead of large dams CBA (B–A)

5031 4700 331 3439 3108

5591 4700 891 3439 2548

6525 4700 1825 3439 1614

9534 4700 4534 3439 − 1095

Cost-benefit analysis RES versus fossil fuels ($ US millions) Cost of fossil fuels scenario (C) Difference between RES and fossil fuels investment costs (D) WTP premium for RES instead of fossil fuels CBA (D–C)

4285 746 4059 3313

4285 1306 4059 2753

4285 2240 4059 1819

4285 5249 4059 − 1190

premium for RES. Households that plan to visit the Aysen Region are willing to pay a higher premium than those who do not, but surprisingly, the opposite effect is found for those who have already visited the area. The knowledge about current electricity generation sources has a positive and significant effect only in the case where fossil fuels are the baseline. This indicates that people who know which energy sources are currently used in Chile will be willing to pay more for RES when they know that the extra energy would otherwise be produced using fossil fuels. This effect is not significant when the baseline alternative is the construction of large dams in Chilean Patagonia. This might indicate an important concern of the respondents regarding the environmental impact of thermoelectric developments. Finally, results obtained in Table 3 are aggregated by the number of households connected to the CIS to calculate the annual households' aggregate WTP for RES. This value can be compared with the cost of RES projects to conduct a cost-benefit analysis. Table 4 shows the comparison between the calculated WTP premium for RES and the cost of four different future RES scenarios for Chile. The analysis is based on the costs of real RES projects presented to the Foment Corporation in Chile in 2008 (see InvestChile Corfo Project Directory, 2008). To investigate whether the household price premium would exceed the investment cost, the estimated WTP was aggregated by the number of households in the CIS, 4,033,471, and converted to a yearly base. The scenario presented to the respondents in the CV questionnaire was designed to provide an additional 2500 MW of RES within 20 years. For this reason the present value of the monetary flows were calculated using the Chilean social discount rate, of 8% (MIDEPLAN, 2009). We consider four different scenarios for a mix of RES. Scenario I shows a plan based on the official RES project mix for 2008, which consists of 90% wind power, 9% biomass and 1% solar power (InvestChile Corfo Project Directory, 2008). Scenario II consists of 75% wind power, 15% biomass and 10%

solar power, and Scenario III consists of 50% wind power, 25% biomass and 25% solar power. Although the scenario presented to respondents in the questionnaire did not consider geothermal power, from Table 2 it is possible to note that 20% of households support this source. Thus, assuming constant preferences related to the other sources, a fourth scenario including geothermal power is also analyzed. This scenario is composed of 80% wind, 9% biomass, 1% solar and 10% geothermal power. Results show that the premiums that households are willing to pay to develop RES are larger than the additional costs of developing these sources over the costs of developing either large dams or fossil fuels energy in all scenarios, except in Scenario IV, which includes investments in very costly geothermal energy. For example, comparing scenario I with the large dam scenario in Chilean Patagonia, we find that even if the costs of RES are larger than the costs of developing the large dams by 331 million USD, the aggregate additional benefits that people receive from developing RES are 3439 million USD, thus indicating that the additional benefits from developing RES are greater than the additional costs for scenario I by 3108 million USD. 7.3. Results on advance disclosure learning and internal consistency of SB-DB data In order to test our methodological hypothesis on the effect of introduction of advanced disclosure learning design in a DBDC contingent valuation study we estimated the single and double bounded models separately and calculated their respective mean WTP and differences (WTPSB − WTPDB). The models in this section were estimated using the variable price (bid) as the only explanatory variable. Table 5 shows results of the estimations and the mean WTP for each elicitation format. As expected, we find the negative and significant sign for the bid vector. We applied a bootstrap procedure with 10,000 replications as explained in section five, to test the hypothesis

Table 5 Results single and double bounded dichotomous choice logit models for methodological test of DBDC internal consistency (t-values in parentheses). RES instead of hydropower

RES instead of fossil fuels

Single bounded

Double bounded

Single bounded

Double bounded

Variable

Coefficient

Coefficient

Coefficient

Coefficient

Constant Bid Mean WTP WTP St. error Log-likelihood function Sample size

2.50 (11.53) − 0.70 (− 6.84) 3590 285.50 − 319.49 663

2.35 (19.579) − 0.70 (− 18.00) 3347 130.40 − 745.02 663

1.90 (10.89) − 0.46 (− 10.11) 4109 204.23 − 389.02 663

1.67 (16.28) − 0.42 (− 18.49) 3982 185.79 − 720.84 663

C. Aravena et al. / Energy Economics 34 (2012) 1214–1225

1500 1000

Frequency

500

0

0

500

Frequency

RES Instead of Fossil Fuels

1000 1500 2000 2500 3000

RES Instead of Hydropower

1223

-500

0

500

1000

1500

2000

-200

WTP Differences in Chilean Pesos (CLP)

E(WTP)SB –E(WTP)DB :

245 CLP (-142.72 to +906.18)

0

200

400

600

WTP Differences in Chilean Pesos (CLP)

E(WTP)SB - E(WTP)DB :

Ho: Reject

117 CLP (-103.74 to +365.19)

Ho: Reject

Fig. 3. Internal consistency test for the difference between SB-DB WTP. Ho: WTPSB − WTPDB ≠ 0 (95% confidence intervals in parenthesis).

of significant differences between single and double bounded welfare estimates (see hypothesis tests in Eqs. (13) and (14)). 21 Results are presented in Fig. 3. Results of the both hypothesis tests show no significant differences in SB and DB estimates of mean WTP. For the first time in the literature we find that the introduction of advanced disclosure in the DB mechanism provides consistency between SBDB WTP estimates, adding credibility to this preferred CVM format. Furthermore, the follow up valuation of the second good also shows closer internal consistency in terms of decreasing differences in SB and DB WTP, evidencing the presence of learning as previously found in Bateman et al., 2008.

8. Conclusion This paper studies the preferences of Chilean households for additional future electricity supplies from different energy sources – fossil fuels, large hydropower in Chilean Patagonia and renewable energy sources – and considers their associated environmental and social impacts. By using the Contingent Valuation method we elicited the willingness to pay premiums for the future introduction of RES, instead of increasing the use of fossil fuels or developing large dams in Chilean Patagonia, in order to supply the increasing electricity demands of the country. Results show that Chilean households support the introduction and development of RES by showing positive and significant WTP premiums in order to avoid the environmental and social impacts associated with each of the baseline alternative sources, which in the case of this study are fossil fuels and large dams in Chilean Patagonia. This indicates that RES are the preferred sources of electricity by

21 We apply the bootstrap procedure also in this part due to the responses to the SB and DB referendum correspond to the same individual, generating samples that are non-independent.

households over the other alternatives available to them for future electricity generation. This situation shows that the public have a clear concern about the environmental and social externalities associated with energy generation and they have a clear preference for less environmentally damaging energy sources. Households impose significant environmental price penalties on energy sources seen as environmentally damaging. The price premiums for RES represent an approximation to the value of the externalities arising from energy generation by the other two alternative sources. These externalities take the form of well documented environmentally and socially damaging impacts. These environmental prices are an important issue to be considered when evaluating projects and making decisions on selection between different electricity sources. From our results, at the aggregate level (considering Chilean households connected to the Central Interconnected Electrical System), the total annual price premium for RES rises to $350 million USD (164,614 million CLP) over hydropower from large dams in Chilean Patagonia and $413 millions USD (194,284 million CLP) over electricity from fossil fuels. 22 These figures can be considered as environmental prices that the utilities in charge of the projects should include in cost-benefit analysis in order to internalize the associated cost and to decide on future electricity investment. By using a pair-wise bootstrap to test for differences in the RES price premiums for the two scenarios under analysis we found that the WTP premium for RES when thermoelectric sources are the baseline is significantly higher than the premium for RES when the baseline is building large dams in Chilean Patagonia. This indicates that the environmental and social impacts caused by fossil fuels – air pollution, contribution to climate change, and energy dependency – are considered of a higher value than impacts from building large dams in Chilean Patagonia. This result is also interesting as it allows

22

Values are in nominal terms for 2008.

1224

C. Aravena et al. / Energy Economics 34 (2012) 1214–1225

establishing rankings for projects developments, which would consider the public concerns and would increase social acceptance. The positive WTP or environmental premiums for RES appear to be enough in aggregated terms to cover the officially reported extra investment cost needed to develop RES projects based on wind, solar and biomass over conventional electricity sources. The premiums are not found to be large enough to exceed the high additional costs of geothermal sources, over conventional sources however. The Chilean authorities should take into account these environmental prices and consider the diversification of the energy mix. They could consider the implementation of “green tariff schemes” or “price premiums for green electricity”, as established in most European countries. To support the development of RES the government could provide a subsidy equal to about 16% to 19% of the price of electricity, as this is the amount that on average the respondents in our study are willing to pay to support RES. Whether such a subsidy should be provided in the form of a green tradable certificate, a feed-in tariff or other policy instruments is beyond the objective of this study. Further research is needed to investigate the most appropriate policy instruments to be introduced in developing countries to support RES. Some theoretical studies (Lintunen and Kangas, 2009; Mulder, 2008) discuss alternative policy instruments in a European context, (e.g. feed in tariffs or subsidies). These studies could be used in the design of energy policy for countries adopting new technology. This study has shown that it is important in the assessment process of electricity projects to account for the externalities produced by the different means of generation, through the introduction of environmental pricing in cost-benefit analysis. From the case study in this research, it is noticeable that the values attached by households to environmental and social impacts of traditional energy generation sources are significant amounts. The introduction of these environmental prices into cost-benefit analysis clearly shows that RES would become a more competitive source. This should be evaluated case by case for the different energy projects presented in the future, and valuation techniques can become a helpful tool in this task. Finally, we show that learning by advanced disclosure about the double referendum mechanism can produce internally consistent welfare measures by reducing the difference between single bounded and double bounded estimates in contingent valuation. This validity check is a feature of major importance to be considered in all future designs of double bounded dichotomous choice contingent valuation studies as it overcomes the major objection to the validity of this preferred CVM format. Acknowledgments The authors are grateful to the Latin American Environmental Economics Program (LACEEP) and the Gibson Institute at Queens University Belfast for financial support. The authors are also grateful to Professor Fredrik Carlsson, Mr. Dave Matthews, Dr. Juan P. Orrego, Juan Riquelme and the participants to the LACEEP workshops and the 10th International Association of Energy Economics European Conference for helpful comments on previous versions of this paper. The usual disclaimer applies. Appendix A. Supplementary data Supplementary data to this article can be found online at doi:10. 1016/j.eneco.2011.11.004. References Aadland, D., Caplan, A., 2006. Cheap talk considered: new evidence from CVM. J. Econ. Behav. Organ. 60, 562–578. Alberini, A., 1995. Efficiency vs bias of willingness-to-pay estimates: bivariate and interval-data models. J. Environ. Econ. Manag. 29, 169–180.

Alberini, A., Kanninen, B., Carson, R.T., 1997. Modeling response incentive effects in dichotomous choice contingent valuation data. Land Econ. 73, 309–324. Álvarez-Farizo, B., Hanley, N., 2002. Using conjoint analysis to quantify public preferences over the environmental impacts of wind farms. An example from Spain. Energy Policy 30, 107–116. Arrow, K., Solow, R., Portney, P.R., Leamer, E.E., Radner, R., Schuman, H., 1993. Report of the NOAA panel on contingent valuation. Fed. Regist. 58, 4601–4614. Bateman, I.J., Willis, K.G., 1999. Valuing Environmental Preferences: Theory and Practice of the Contingent Valuation Method in the US, EU, and Developing Countries. Oxford University Press, USA. Bateman, I.J., Langford, I.H., Jones, A.P., Kerr, G.N., 2001. Bound and path effects in double and triple bounded dichotomous choice contingent valuation. Resource Energy Econ. 23, 191–213. Bateman, I.J., Carson, R.T., Day, B., Hanemann, M., Hanley, N., Hett, T., Jones-Lee, M., Loomes, G., Mourato, S., Ozdemiroglu, E., Pearce, D.W., Sugden, R., Swanson, J., 2002. Economic Valuation with Stated Preference Techniques: a Manual. Edward Elgar, Ltd, Cheltenham. Bateman, I.J., Cole, M., Cooper, P., Georgiou, S., Hadley, D., Poe, G.L., 2004. On visible choice sets and scope sensitivity. J. Environ. Econ. Manag. 47, 71–93. Bateman, I.J., Burgess, D., Hutchinson, W.G., Matthews, D.I., 2008. Learning design contingent valuation (LDCV): NOAA guidelines, preference learning and coherent arbitrariness. J. Environ. Econ. Manag. 55, 127–141. Batley, S., Fleming, P., Urwin, P., 2000. Willingness to pay for renewable energy: implications for UK green tariff offering. Indoor Built Environ. 9, 157–170. Batley, S., Colbourne, D., Fleming, P., Urwin, P., 2001. Citizen versus consumer: challenges in the UK green power market. Energy Policy 29, 479–487. Bergmann, A., Hanley, N., Wright, R., 2006. Valuing the attributes of renewable energy investments. Energy Policy 34, 1004–1014. Bollino, C.A., 2009. The willingness to pay for renewable energy sources: the case of Italy with socio-demographic determinants. Energy J. 30, 81–96. Borchers, A.M., Duke, J.M., Parsons, G.R., 2007. Does willingness to pay for green energy differ by source? Energy Policy 35, 3327–3334. Burton, A.C., Carson, K.S., Chilton, S.M., Hutchinson, W.G., 2003. An experimental investigation of explanations for inconsistencies in responses to second offers in double referenda. J. Environ. Econ. Manag. 46, 472–489. Burton, A.C., Carson, K.S., Chilton, S.M., Hutchinson, W.G., 2009. Why do people nondemand revel in hypothetical double referenda for public goods. Appl. Econ. 41, 3561–3569. Byrnes, B., Jones, C., Goodman, S., 1999. Contingent valuation and real economic commitments: evidence from electric utility green pricing programmes. J. Environ. Plan. Manag. 42, 149–166. Cameron, T.A., Quiggin, J., 1994. Estimation using contingent valuation data from a “dichotomous choice with follow-up” questionnaire. J. Environ. Econ. Manag. 27, 218–234. Campbell, D., Aravena, C., Hutchinson, W.G., 2011. Cheap and expensive alternatives in stated choice experiments: are they equally considered by respondents? Appl. Econ. Lett. 18, 743–747. Carlsson, F., Martinsson, P., 2006. Do Experience and Cheap Talk influence Willingness to Pay in an Open-Ended Contingent Valuation Survey? Working Papers in Economics 190. Göteborg University, Department of Economics. Carson, R.T., Groves, T., 2007. Incentive and informational properties of preference questions. Environ. Resour. Econ. 37, 181–210. Carson, R.T., Hanemann, W.M., Mitchell, R.C., 1986. Determining the Demand for Public Goods by Simulating Rederendums at Different Tax Prices. Working Paper. Department of Economics. University of California, San Diego. Carson, R.T., Mitchell, R.C., Hanemann, M., Kopp, R.J., Presser, S., Ruud, P.A., 2003. Contingent valuation and lost passive use: damages from the Exxon Valdez oil spill. Environ. Resour. Econ. 25, 257–286. Chilean General Law of Electric Services Number 20257, 2008. Chilean General Energy Law of Electric Services. , p. 7. Cooper, J.C., Hanemann, M., Signorello, G., 2002. One-and-one-half-bound dichotomous choice contingent valuation. Rev. Econ. Stat. 84, 742–750. Cummings, R., Taylor, L., 1999. Unbiased value estimates for environmental goods: a cheap talk design for the contingent valuation method. Am. Econ. Rev. 89, 649–665. Cummings, R.G., Brookshire, D.S., Schulze, W.D., 1986. Valuing Environmental Goods: an Assessment of the Contingent-Valuation Method. Rowan and Allenheld, Totowa, NJ. DeShazo, J.R., 2002. Designing transactions without framing effects in iterative question formats. J. Environ. Econ. Manag. 43, 360–385. Diaz-Rainey, I., Ashton, J., 2007. Characteristics of UK Consumers' Willingness to Pay for Green Energy. Working Paper. Social Sciences Research Network. Efron, B., Tibshirani, R.J., 1993. An Introduction to the Bootstrap. Chapman & Hall/CRC, London. Ek, K., 2002 Valuing the environmental impacts of wind energy: A choice experiment approach. Licentiate thesis. Department of Business Administration and Social Sciences. Lulea University of Technology. October. Sweden. European Commission, 2003. External Costs: Research Results on Socio-environmental Damages due to Electricity and Transport. Directorate-General for Research, Directorate J-Energy, Brussels. European Commission, 2005. ExternE — Externalities of Energy: Methodology 2005 Update. Directorate-General for Research, Sustainable Energy Systems, Brussels. Färe, R., Grosskopf, S., Pasurka, C.A., 2010. Toxic releases: an environmental performance index for coal-fired power plants. Energy Econ. 32, 158–165. Farhar, B.C., 1999. Willingness to Pay for Electricity from Renewable Resources: a Review of Utility Market Research. NREL/TP- 550-26148. National Renewable Energy Laboratory, Golden, CO. July.

C. Aravena et al. / Energy Economics 34 (2012) 1214–1225 Farhar, B.C., Coburn, T., 1999. Colorado Homeowner Preferences on Energy and Environmental Policy. NREL/TP-550-25285. National Renewable Energy Laboratory, Golden, CO. June. Farhar, B.C., Houston, A.H., 1996. Willingness to Pay for Electricity from Renewable Energy NREL Technical Report. National Renewable Energy Laboratory, Golden, CO. Georgakellos, D., 2010. Impact of a possible environmental externalities internalization on energy prices: the case of the greenhouse gases from the Greek electricity sector. Energy Econ. 32, 202–209. Groothuis, P.A., Groothuisand John, C., Jana, D., 2008. Green vs. green: measuring the compensation required to site electrical generation windmills in a viewshed. Energy Policy 36, 1545–1550. Haab, T.C., McConnell, K.E., 2002. Valuing Environmental and Natural Resources: the Econometrics of Non-market Valuation. Edward Elgar Publishing. Hall, S., Román, R., Cuevas, P., 2009. ¿Se necesitan represas en La Patagonia? Un análisis del futuro energético chileno. Ocholibros, Universidad de Chile, Chile. Hanemann, W.M., 1984. Welfare evaluations in contingent valuation experiments with discrete responses. Am. J. Agric. Econ. 66, 332. Hanemann, M., Loomis, J., Kanninen, B., 1991. Statistical efficiency of double-bounded dichotomous choice contingent valuation. Am. J. Agric. Econ. 73, 1255. Hanley, N., Nevin, C., 1999. Appraising renewable energy developments in remote communities: the case of the North Assynt Estate, Scotland. Energy Policy 27, 527–547. Hutchinson, W.G., Scarpa, R., Chilton, S.M., McCallion, T., 2001. Parametric and Nonparametric estimates of willingness to pay for forest recreation in Northern Ireland: a discrete choice contingent valuation study with follow-ups. J. Agric. Econ. 52, 104–122. InvestChile Corfo Project Directory, 2008. Renewables and SDM in Chile. Investment Opportunities and Project Financing. Document of the Chilean Economic Development Agency. IPCC, Climate Change, 2007. Synthesis Report. In: Core Writing Team, Pachauri, R.K., Reisinger, A. (Eds.), Contribution of Working Groups I, II and III to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. IPCC, Geneva, Switzerland. 104 pp. Ivanova, G., 2005. Queensland Consumers' Willingness to Pay for Electricity from Renewable Energy Sources. Working paper of the Australia New Zealand Society for Ecological Economics. Paper presented in the Ecological Economics in Action Conference in Massey University, Palmerston North, New Zealand, December 11–12, 2005. Available at: http://www.anzsee.org/anzsee2005papers/Ivanova_%20WTP_ for_renewable_energy.pdf. Kanninen, B.J., 1993. Optimal experimental design for double-bounded dichotomous choice contingent valuation. Land Econ. 69, 138–146. Kanninen, B.J., 1995. Bias in discrete response contingent valuation. J. Environ. Econ. Manag. 28, 114–125. Koundouri, P., Kountouris, Y., Remoundou, K., 2009. Valuing a wind farm construction: a contingent valuation study in Greece. Energy Policy 37, 1939–1944. Lintunen, J., Kangas, H.L., 2009. The case of co-firing: the market level effects of subsidizing biomass co-combustion. Energy Econ. 32, 694–701. Longo, A., Markandya, A., Petrucci, M., 2008. The internalization of externalities in the production of electricity: willingness to pay for the attributes of a policy for renewable energy. Ecol. Econ. 67, 140–152. McFadden, D., 1974. Conditional logit analysis of qualitative choice behavior. Front. econometrics 8, 105–142.

1225

McFadden, D., 1994. Contingent valuation and social choice. Am. J. Agric. Econ. 76, 689–708. Meyerhoff, J., Ohl, C., Hartje, V., 2010. Landscape externalities from onshore wind power. Energy Policy 38, 82–92. MIDEPLAN, 2006. División Social, National survey data CASEN — the National Socioeconomic Characterization Survey. Available at: www.mideplan.cl/casen. MIDEPLAN, 2009. Precios Sociales para la Evaluación Social de Proyectos. Chilean Government. Available at www.mideplan.cl. Miller, G.T., Spoolman, S.E., 2009. Sustaining the Earth, Ninth ed. Cengage Advantage Books, Belmont, Canada. Mitchell, R.C., Carson, R.T., 1989. Using Surveys to Value Public Goods: the Contingent Valuation Method. Resources for the Future, Washington. Mulder, A., 2008. Do economic instruments matter? Wind turbine investments in the EU (15). Energy Econ. 30, 2980–2991. NEC Report on electricity prices. April 2007. Available at www.cne.cl and Statistics of CDEC and Centro de Despacho Económico de Carga (Economic Center of Charge Service). Available at: http://www.cdec-sic.cl/ Nomura, N., Akai, M., 2004. Willingness to pay for green electricity in Japan as estimated through contingent valuation method. Appl. Energy 78, 453–463. Operation Statistics, 1999/2008. CDEC-SIC. Central Interconnected System/Load Economic Dispatch Center. Also available at: http://www.cdec-sic.cl/datos/anuario 2009/imagenes/cdec_ing.pdf. Owen, A.D., 2006. Evaluating the cost and benefits of renewable energy technologies. Aust. Econ. Rev. 39, 207–215. Poe, G., Welsh, M., Champ, P., 1997. Measuring the difference in mean willingness to pay when dichotomous choice contingent valuation responses are not independent. Land Econ. 73, 255–267. Polinori, P., 2009. Italians, Renewable Energy Sources and EU “Climate Vision.” IAEE Newsletter, Second Quarter. Roe, B., Teisl, M.F., Levy, A., Russell, M., 2001. US consumers' willingness to pay for green electricity. Energy Policy 29, 917–925. Rowlands, I., Scott, D., Parker, P., 2003. Bus. Strateg. Environ. 12, 36–48. Scarpa, R., Bateman, I., 2000. Efficiency gains afforded by improved bid design versus follow-up valuation questions in discrete-choice CV studies. Land Econ. 76, 299–311. Scarpa, R., Willis, K., 2010. Willingness-to-pay for renewable energy: primary and discretionary choice of British households' for micro-generation technologies. Energy Econ. 32, 129–136. Sundqvist, T., 2002. Power Generation Choice in the Presence of Environmental Externalities. Department of Business Administration and Social Sciences, Lulea University of Technology, Sweden. Verbeet, B., 2007. The willingness to pay for green electricity. AENORM 56, 13–16 August. Whitehead, J.C., Cherry, T.L., 2007. Willingness to pay for a Green Energy program: a comparison of ex-ante and ex-post hypothetical bias mitigation approaches. Resour Energy Econ. 29, 247–261. Wiser, R.H., 2007. Using contingent valuation to explore willingness to pay for renewable energy: a comparison of collective and voluntary payment vehicles. Ecol. Econ. 62, 419–432. Zarnikau, J., 2003. Consumer demand for green power and energy efficiency. Energy Policy 31, 1661–1672.