CAIRN.INFO : Matières à réflexion


1In France, hydropower production is subject to a concession. The state grants private operators the possibility to withdraw water for electricity generation. In the coming years, a significant share of such concessions will expire and they will be re-assigned through a competitive public procedure. The renewal procedure requires petitioners to present three offers:

  • A technical offer, whereby bidders are expected to improve the existing infrastructure to increase electricity production (if possible);
  • An environmental offer, whereby bidders must show their plans to reduce environmental impacts;
  • An economic offer, whereby bidders have to offer a percentage of their revenues as a royalty payment to the state.

2In accordance with the Water Framework Directive (WFD, 2000/60/EC) and in order to avoid an infringement procedure, the French Government has to take concrete actions so that rivers attain a “good ecological status” in the coming years. Given the situation, the environmental aspect of the renewal procedure is considered one of the top actions to be carried out.

3The impact of hydropower projects on the environment varies greatly, depending on site-specific mitigation measures and production strategies: if badly managed, hydropower production can reduce biodiversity and can significantly degrade fluvial ecosystems and associated ecosystem services (among others: Céréghino, Cugny and Lavandier [2002]; Croze, Bau and Delmouly [2008]; Brown, Hannah and Milner [2009]; Renofalt, Jansson and Nilsson [2010]; Halkos and Matsiori [2014]).

4Therefore, the Government is expecting bidders to come up with effective mitigation measures, ranging from fish-passages to complex outflow reservoirs aimed at minimizing flow changes generated by hydro-peaking. As bidders have to submit both an economic offer and an environmental one, it is likely that offers with higher environmental improvements will be coupled with offers having a lower royalty percentage.

5The scope of our paper is to analyse the emerging trade-off between a healthier environment and more revenues for local authorities. To do so we studied the Aspe River. We do not consider the technical offer as hydropower technology is mature, technical improvements are limited (particularly for existing plants) and technical options and solutions are fairly standard (International Renewable Energy Agency [2012]).

6Sixteen hydropower plants on the Aspe River (Figure 1), for a total installed capacity of 93 megawatts (MW), are among the concessions that will be subject to the renewal process. The renewal procedure is designed so that the winner will operate all sixteen power plants; therefore, after the renewal, the hydropower plants will be managed as a single hydropower scheme.

Figure 1

Localization of the Aspe River

Figure 1

Localization of the Aspe River

7The Aspe is a torrential river flowing through the Aspe Valley, one of the three main valleys of the High Béarn in the Southwest of France. The Aspe River is part of Natura 2000, an ecological network of protected areas within the European Union. The Aspe is one of the few French rivers where salmons migrate for reproduction and some measures to facilitate their migration were put in place. Still, due to the artificial hydromorphology, the success rate of downstream migration of salmon smolts is 55% (Chanseau, Larinier and Travade [1999]).

8The specificity of our work is the selection of the bidding vehicle in the discrete choice experiment (DCE), whereby we translate the revenue sharing percentage to an immediate rebate on electricity bills. Monetary valuations are often elicited in environmental valuations. The resulting willingness to pay (WTP) or to accept (WTA) is a measure of how much the respondent values the intervention. Following Hicks [1943], one can use either the concept of compensating variation or the concept of equivalent variation to measure monetary values. Compensating variation measures the amount of money that is required after the change to make a respondent’s level of utility the same as before the change, while equivalent variation measures the amount of money that is required before the change, to make utility the same as it would be after the change. Within both these concepts, a distinction can be made between WTA (when compensation is required) and WTP (when a payment is required). In our experiment, potential gainers from the environmental improvement are asked to state their maximum WTP for the intervention, with different attributes representing the unit change of several dimensions included in the intervention. Therefore, our DCE is coherent with a willingness to pay approach (WTP), with the advantage of avoiding the use of a tax, which entails potential negative bias in respondents (Morrison, Blamey and Bennett [2000]).

9In reality, the law does not foresee a direct rebate in the bills. However, the beneficial effect of an electricity bill rebate has similar benefits as the reduction in local taxes or an improvement in local services, made possible by the revenues accruing from the bidding process.

10We adopt two approaches: 1) the standard approach in preference space, in which the distributions of the coefficients are estimated and WTP is derived from the ratio of two coefficients; and 2) the WTP space approach, in which we specify the distribution of WTP directly at the estimation stage. To the best of our knowledge, few studies have used the WTP space specification in DCE analysis, especially in environmental economics (Train and Weeks [2005]; Scarpa, Thiene and Train [2008]; Balcombe et al. [2008]; Balcombe, Chalak and Fraser [2009]). Hole and Kolstad [2012] have recently compared these two methods in health economics and found that the preference space approach fits the data better but the distribution of WTP is more realistic in the WTP space.

11This paper shows that people are willing to pay to increase the ecological status of the Aspe River; the highest total WTP is approximately €1,225 per household per year. The highest marginal WTP for a satisfactory fish stock reaches €277 per year, three times the maximum rebate offered. Having restricted the surveyed population to the locals, while reinforcing the coherence of the study, explains these relatively high values. These values remain acceptable, if compared for instance to the results obtained by Kataria [2009] in a DCE study extended to all Swedish rivers.

12Finally, we compare our results to the restoration costs incurred for the cleanup of the Aspe River in the aftermath of a major environmental accident, which took place in 2007. This gives a more concrete assessment of the aggregate WTP order of magnitude.

13The paper unfolds as follows: in the second section, we review the literature on aquatic ecosystem evaluation; in the third section, we discuss how we have structured the bidding vehicle for the DCE; in the fourth section, we describe the experimental design and econometric approach; in the fifth section, we present the results of the choice experiment; and in the sixth section, we offer conclusions.

Literature review on aquatic ecosystem evaluation

14There are several ways to monetize environmental impacts. It is beyond the scope of this paper to discuss the pros and cons of each methodology (for a critical assessment, see Bateman et al. [2002]). Given the multidimensional and complex nature of ecosystems, there is ample scientific consensus that the best method for estimating how a combination of changes to one or more ecosystem services affects human welfare is the DCE (Hoyos [2010]).

15Concerning environmental goods, and fluvial ecosystems in particular, it is important to relate a change in attribute levels to an action, typically a change in policy or a change in management of the resource. A standard procedure when using DCE for environmental goods is to include in every choice set an alternative that reflects either the current status (status quo) of the good being evaluated or an opt-out alternative (i.e., the worst possible situation). Normally, the price (or cost) of these alternatives is set equal to 0. The DCE format allows one to easily convert marginal utility estimates for changes in the level of each attribute to WTP estimates. Moreover, given that compensating variation measures may be obtained, it is possible to estimate the total value of improvements to the environmental good as a consequence of the policy or managerial change (Crastes and Mahieu [2014]).

16Crucial elements for the design of DCE relating to environmental impacts in water bodies include the following: the definition of the affected population, the delimitation of the water bodies being studied, and the attributes selected to describe the environment.

17The study population may range from only individuals who use or reside near the water bodies being studied (Hynes, Hanley and Scarpa [2008]; Kataria et al. [2012]; Stithou et al. [2012]) to a representative sample of the regional or national population (Kataria [2009]; Metcalfe et al. [2012]). The target population clearly depends on both the expected effects of the policy (or managerial changes under consideration) and on the water bodies under consideration (Hanley, Wright and Alvarez-Farizo [2006]; Brouwer, Martin-Ortega and Berbel [2010]; Poirier and Fleuret [2010]; Kataria [2009]; Metcalfe et al. [2012]).

18Normally, attributes used in DCE surveys relate the ecology of a water body to recreational opportunities and to the aesthetics of the environment. A successful choice experiment requires testing of the attributes relevant to stakeholders, typically meaning the public. Consequently, the attributes, or the levels used in the questionnaires, must be linked to the environmental attributes used to assess the impacts but need not to be the same. Consider the following simplified example: the attribute water quality can be expressed in terms of its different levels of chemical components or in simpler terms, such as a swimmable or non-swimmable river; it is straightforward that this attribute, familiar to the public, depends on the level of certain chemicals. Thus, attribute levels are commonly qualitative (Hanley, Adamowicz and Wright [2005]; Alvarez-Farizo et al. [2007]; Birol, Koundouri and Kountouris [2008]) and sometimes use images or visual descriptions (Doerthy, Campbell and Hynes [2013]). The most common attributes include the following: biodiversity, generally described as various quantities of native species (Morrison and Bennett [2004]; Kragt et al. [2011]); recreational activities (Doerthy, Campbell and Hynes [2013]); and the quality of aesthetics (Alvarez-Farizo et al. [2007]; Hanley, Wright and Alvarez-Farizo [2006]; Stithou et al. [2012]).

19To our knowledge, only one paper (Kataria [2009]) used DCE to estimate how individuals value different environmental improvements for rivers where hydropower production takes place. This paper focuses on all Swedish rivers and its aim is to assess the market share of environmentally friendly producers, which are expected to face higher production costs. According to Kataria’s analysis, the total willingness to pay for the single attributes range between SEK1,100 and SEK1,400 (€119 and €211 respectively). The maximal willingness to pay for a bundle of different attributes was estimated to SEK2,100 (€226). With respect to this paper, we innovate with regard to both the econometric methodology, as explained above, and the treatment of the bidding vehicle, as detailed below.

Bidding vehicle and sample selection

20The peculiar aspect of our DCE relates to the bidding vehicle. As we wanted to avoid the use of a tax as a bidding vehicle for acceptability reasons (Morrison, Blamey and Bennett [2000]), we preferred to opt for a rebate on electricity bills, which is normally associated with WTA analyses. To our knowledge, only once it was used as a WTP (Burton and Rigby [2009]).

21Technically, the value of a benefit is measured by the WTP to secure that benefit and the value of a cost is measured by the WTA compensation for a loss. As stated by Mitchell and Carson [1989] and Pearce, Atkinson and Mourato [2006], even with regard to environmental goods, individuals have a property right to the status quo, and the relevant value in this context is their WTA compensation for its loss. However, individuals have no right to the benefit brought about by the policy (or measure) in question, and the relevant value is therefore their WTP.

22The renewal procedure requires bidders to offer a royalty as well as improvements to the fluvial ecosystem. From an environmental perspective, this means either the fluvial ecosystem will remain at the status quo or it will improve. Either in addition, the winner will share the current percentage of revenue with central and local authorities. Bidders will therefore offer various combinations of environmental improvement and revenue sharing. Both strategies have minimum thresholds: the health of the ecosystem cannot fall below the status quo and the royalty cannot be less than 0%.

23In summary, regardless of the specific combination, the existing situation for local communities will not worsen. Thus, our questionnaire is designed to investigate whether people prefer higher levels of ecosystem mitigation or higher percentages revenue sharing with local authorities. Whenever a trade-off emerges, it is important to test people’s preferences. To do so effectively, it is essential to develop an understandable way of communicating the trade-off. In this case, we have imagined that the revenue sharing percentage can be translated into immediate rebates on electricity bills. Although, in reality, there will be no rebate, the revenue sharing would increase revenue for local authorities, which should result in lower local taxes or improved local services.

24Because improving fluvial ecosystems is costly, we expect higher levels of ecosystem recovery to be associated with lower offers for revenue sharing; conversely, higher offers for revenue sharing are expected to come at the price of lower levels of ecosystem mitigation. This led us to devise an opt-out alternative comprised of the status quo for ecosystem health and the highest possible rebate. It is important to stress that the opt-out alternative is not the current status quo; it is a hypothetical situation, in which individuals receive the highest rebate on their electricity bill. All other situations leave them with less money but a healthier ecosystem. Therefore, the experiment adopts a WTP approach. We ask participants whether they are willing to renounce money they could spend on something else in favor of a healthier fluvial ecosystem.

25The touristic flow is very limited in the Aspe Valley, which is a small area. Most importantly, only the local population is concerned by a potential change of the river ecosystem, on one side, and the compensation offered by an hydroelectric concession bidder. Therefore, as the benefits will accrue to local authorities, we decided to address the choice experiment only to individuals living in the region. Although this choice restricts the geographical scope of the analysis, it offers a coherent scope to our purpose. [1]

Experimental design: structure, attributes and levels

26Our questionnaire had two parts. In the first part, respondents were asked questions about their attitude towards the Aspe River and their socio-economic status. The second part contained the choice experiment.

27Attributes and levels relevant for the Aspe River ecosystem were chosen using a Delphi survey coordinated by the local water agency (Agence de l’eau Adour-Garonne), which involved 15 selected experts. The Delphi survey was critical not only for defining the attributes and their levels but also for confirming that various hydropower management regimes effectively increase the quality of the fluvial ecosystem. The Delphi survey revealed three attributes that are most relevant for the Aspe ecosystem, namely water quality, fish population and hydromorphology. Moreover, the Delphi survey made it possible to define the current state of these three attributes describing the fluvial ecosystem. As discussed in the section “Restoration costs and ecosystem value,” in 2007, a major environmental accident affected the Aspe ecosystem. The river recovered its pre-accident status in 2011 (MEDDTL [2011]). During the recovery period, local communities experimented first-hand both the implications of lower water quality and the positive environmental effects of a more natural hydromorphology. Agency directors confirmed that the attributes chosen were relevant and understandable attributes for the local population. To aid understanding, all attribute levels have been expressed in qualitative and figurative terms. Finally, experts provided images and visual descriptions of the attributes described.

28Water quality represents the chemical and physical conditions of the water. The attribute is represented qualitatively, according to the scale provided by the water agency and to which local population got used to, during the recovery period.

29The second attribute is fish population. Hydropower production normally has a consistent negative impact on the natural reproduction of fish (Renofalt, Jansson and Nilsson [2010]). The Aspe River is one of the last rivers in the Pyrenees where Atlantic salmon and the sea trout migrate to reproduce (COGEPOMI [2008]). The protection of these species is crucial and these species are essential elements of the Aspe ecosystem. The levels chosen were qualitative according to the scale provided by the Aquitaine Regional Environmental Agency (direction régionale de l’Environnement Aquitaine, DRE Aquitaine).

30The third attribute is hydromorphology, which indicates whether a river has a natural flow. The attribute was represented with images taken from the Aspe River. We used this attribute to determine the number of respondents that value a naturally flowing water body. It is possible for hydropower plants, if properly designed, built and managed, to avoid significantly altering the natural flow of a river, thereby increasing the health of the ecosystem.

Table 1

Attribute and attribute levels

Water QualityChemical conditionsSufficient; Good; Very good.
Fish PopulationAbundance and evolution of the stockUnsatisfactory; Satisfactory.
HydromorphologyCloseness to natural conditionsNatural; Artificial.
RebateReduction of electricity bill per household (in EUR)0; 10; 45; 75.

Attribute and attribute levels

31The maximum rebate was determined taking into account the amount that could accrue to a single household. So far, the only concession that was renewed is located on the Rhone and it was awarded to the Compagnie Nationale du Rhône (CNR) [2013]. The CNR offered a 25% royalty on its revenues. We used this percentage to compute the maximum rebate for each household, which stands at €75, or 15% of the average electricity bill (Commission de Régulation de l’Énergie [2013]). The result was estimated by taking into account:

  • The average electricity price on the Power Exchange for 2013—approximately €50 per MWh (Compagnie Nationale du Rhône [2013]);
  • The fact that 75% of the royalty will accrue to local authorities, while the rest to the central government (Code de l’énergie);
  • The number of households in the Aspe Region, equal to 11,500 (INSEE [2013]).

32A large number of unique hydropower production scenarios can be constructed from this number of attributes and levels. We opted for an unlabeled design and the full factorial of all possible combinations of attributes and levels was reduced to a treatable set of 24 pair-wise comparisons of alternatives with NGene software. The result was a D-efficient design with eight choice sets.

33We labeled each alternative as “electricity supplier x” (with x ranging from 1 to 3), following Kataria [2009]. Thus, in the choice experiment, suppliers differed from each other based on their remedial measures, i.e., the levels of environmental attributes attained. Respondents were then asked to choose their preferred supplier.

Econometric model

34Theoretically, analysis of the choices in DCEs is based on the standard random utility model developed by McFadden [1973], which linked the deterministic model with a statistical model of human behavior as follow:


equation im2

36where αn and βn are individual-specific coefficients for the cost (cnit) and other attributes (xnit); εnit is the error term assumed to be extreme value distributed with variance given by μ2n2 / 6), μn being an individual-specific scale parameter.

37In an econometric perspective, the mixed Logit (MXL) model is the “workhorse” specification for analyzing discrete choice data since it overcomes well-known limitations of the traditional conditional Logit (CL) approach. In brief, the limitations of the CL model are: 1) the Independence of Irrelevant Alternatives (IIA) assumption which states that characteristics of a particular choice alternative do not impact the relative probabilities of choosing other alternatives; 2) restrictive substitution patterns; and 3) absence of preference heterogeneity among respondents (see Train [2003]).

38In practice, most researchers focus on preference space and estimates of the parameters of random coefficients’ distribution using either classical maximum likelihood or Bayesian estimation techniques. The WTP for an attribute is then the ratio of the attribute coefficient to the monetary coefficient. In this context, WTP is the ratio of two randomly distributed terms, which can lead to heavily skewed WTP distributions with no defined moments (see Scarpa, Thiene and Train [2008]; Hole and Kolstad [2012]). To address this potential problem, the monetary coefficient is often specified as fixed or as a constrained distribution.

39According to Train and Weeks [2005], the model in preference space can be specified by dividing the standard random utility model by μn, an individual-specific scale parameter:


equation im3

41where and nit = Unit / μn, λn = αn / μn, ηn = βn / μn, and unit = εnit / μn. This equation implies that the coefficients are independent and therefore random term is assumed to be homoscedastic. In preference space, we implicitly assume that preference heterogeneity is primary main driver that leads individuals to make different choices. However, several papers have suggested that much of the observed preference heterogeneity may be better described as “scale” heterogeneity (see, among others, Louviere, Hensher and Swait [2000] and Train and Weeks [2005]).

42Considering that the WTP for an attribute is given by the ratio ϒn = ηn / λn, Equation (2) can be rewritten as:


equation im4

44Following Train and Weeks [2005], Equation (3) is the model in WTP space. Although both Equations (2) and (3) are equivalent in nature, the WTP specification allows direct specification of the distribution of WTP rather than deriving it from the ratio of two coefficients. The coefficients in the preference space are usually approximated by MXL, and as discussed by Greene and Hensher [2010], the model in WTP space can be expressed as a special case of the generalized multinomial Logit (GMNL) developed by Fiebig et al. [2010]. According to Train [2009], both models can be estimated using either maximum simulated likelihood or Bayesian methods.

45To investigate preferences related to the trade-off between cheaper electricity and a healthier river in the Aspe Valley, we use and compare the two different approaches described above: 1) the standard approach in preference space, in which the distributions of the coefficients are estimated and the WTP is derived from the ratio of two coefficients; and 2) the WTP space approach, in which we specify the distribution of WTP directly at the estimation stage.


46The choice experiment was addressed to households in the Aspe Region (arrondissement Oloron-Saint-Marie). Toluna, a company specialized in web-based surveys, carried out the sampling procedure and programmed the computer-assisted web interviews in order to return at least 100 valid responses from a representative sample. Choice sets were randomized in order to avoid any bias among them. In this study we included a so-called cheap talk script in the questionnaire. Initially suggested by Cummings and Taylor [1999], this is an attempt to bring down the hypothetical bias by thoroughly describing and discussing the propensity of respondents to exaggerate stated WTP. Both Carlsson, Frykblomand and Lagerkvist [2005] have shown the hypothetical bias to be reduced when using a cheap talk script. In the end, we obtained 200 valid responses, for a robust representative sample of 1.7% of all households living in the Aspe area.

47Table 2 reports the demographic statistics for our sample. The mean age of respondents is 41.2 years, and the average number of members in a household is slightly more than 2. Retired or inactive individuals comprise nearly half of the sample. These data are in agreement with demographic statistics from the INSEE and confirm that we have a representative sample. The only significant deviation is the geographical representativeness of our sample: 43% of our respondents live in Oloron-Saint-Marie town, the “capital” of the arrondissement; official statistics indicate that the actual percentage is 12 point lower. Respondents were not previously informed of the relevant characteristics of hydropower production to avoid influencing their choices. Nevertheless, the questionnaire contained concise information on why each attribute was chosen and why it mattered for hydropower production.

Table 2

Descriptive statistics

VariableOur sample (mean)INSEE (2009)
Number of members in household2.222.31
Average income (household)€22,100€22,800
University degree25%21%
Living in Oloron-Saint-Marie town (East and West)0.430.31
Knowledge of concession renewal0.16n.a.
Membership in an environmental organization0.02n.a.

Descriptive statistics

48The utility function of the individual n in choice situation t that we have considered is the following:


equation im5

50where fish2t is the dummy variable for satisfactory level of fish population; hydro2t is the dummy variable for the natural level of hydromorphology; wquality2t and wquality3t instead, are dummies for good and very good level of water quality; and billt, finally, represents the cost increase with respect to the maximum rebate.

51To aid understanding, for all levels of rebates, we have subtracted the maximum level of rebate to create the variable bill. This guarantees that we obtain the standard negative sign for the monetary component of each WTP estimate. All betas represent the marginal utility of each attribute. In the econometric framework we consider models in both preference space and WTP space.

Models in preference space

52In preference space, we consider two Mixed Logit (MXL) models. Model 1 is a random MXL model in which bill and wquality2 attributes are considered to be fixed, and model 2 allows for non-zero correlations between coefficients. For model 2 we assume a completely unrestricted covariance matrix (i.e., no a priori structure for the covariance matrix). For both models the distribution of random coefficients is tested through the semi-nonparametric approach developed by Fosgerau and Bierlaire [2007] comparing normal, log-normal and triangular distributions. Due to space constraints, results of the test are not reported but are available upon request. Each model is estimated based on maximum likelihood using the mixlogit command in Stata (Hole [2007]) assuming a normal approximation. To avoid the problem of local maxima, models are estimated with different starting values. Results are available upon request and are quite similar.

53Table 3 presents estimation results for both models, where all coefficients are significant and have the expected signs. All else equal, the respondents are more willing to preserve the fish population than to improve the water quality or hydromorphology. This finding is not surprising and reflects environmental awareness in the local population. As mentioned above, the Aspe River is one of the last rivers in the Pyrenees where Atlantic salmon and the sea trout migrate to reproduce and, as a result, it is widely believed that recreational activities depend on the presence of these species. Another intriguing result is that people do value the natural flow of the river. As theorized, the environmental accident has increased not only environmental awareness but also knowledge of the river’s hydrology.

Table 3

Results from Mixed Logit (MXL) model in preference space

Model 1: MXLModel 2: MXL (corr)
Non-random parameters
bill– 0.008** (0.003)– 0.007** (0.003)
wquality21.004* (0.140)1.157* (0.143)
Random parameters
fish22.114* (0.205)2.214* (0.196)
hydro20.980* (0.200)1.182* (0.213)
wquality31.093* (0.154)1.323* (0.195)
Standard deviation
hydro21.757* (0.161)1.994* (0.183)
fish21.875* (0.200)1.669* (0.193)
wquality31.389* (0.137)1.717* (0.164)
Log-Lik.– 1323.7– 1279.1

Results from Mixed Logit (MXL) model in preference space

Note: *,** denote rejection at 1% and 5% significance levels. Between parentheses are standard errors. Estimation of model 1 is obtained from 2000 Halton draws simulations, and model 2 from 2500 Halton draws simulations.

54A closer examination of the results of model 1 reveals the estimated standard deviations of several coefficients, with the exception of fish2, are large relative to their mean. This implies that there is a substantial amount of heterogeneity in preferences related to water quality and hydromorphology attributes. We do not observe the same preference heterogeneity for the fish population attribute, confirming the widespread importance of biodiversity among the local population. Results of model 2 are approximately the same except for certain coefficients, which appear to be higher than those in model 1. Overall, model 2 is theoretically more coherent because it allows for correlations between attributes. Indeed, many studies have shown that fish population can be influenced by hydromorphology and water quality (Renofalt, Jansson and Nilsson [2010]).

55The WTP for an attribute is simply the ratio of the distribution of the attribute coefficient and the fixed monetary coefficient. By observing the choices that individuals make when an attribute level changes and also the price associated with this particular scenario of change, we can derive marginal values for each attribute when moving from the opt-out level to each additional level of the attribute, according to the following formula:


equation im6

57where MWTPx,a is the marginal willingness to pay to move from the opt-out level to level a of attribute x; βx,a is the marginal utility of level a of attribute x; β p is the marginal utility of money.

58Table 4 presents the mean, median and standard deviation of the WTP measures derived from models 1 and 2. Regardless of which model is considered, the mean WTP for satisfactory fish population, natural level of hydromorphology, and very good water quality is generally high. As anticipated, the local population is willing to forgo the largest amount of money for a satisfactory fish population: between €250 and €277 (depending on the model) per household per year. Individuals are also willing to pay for natural flow and higher water quality, but hydromorphology of the river appears to be valued least. The distribution of WTP is highly skewed in favor of fish population, with an absolute value of the median that is consistently much higher than the mean. This result indicates that most households are willing to pay most for this attribute relative to other attributes. The WTP distribution for other attributes is somewhat less skewed or even symmetric (see for example wquality3). The standard deviations of the WTP measures are also very large, reflecting the degree of preference heterogeneity among respondents.

Table 4

Willingness to pay for attributes in preference space (€/year)

Model 1Model 2

Willingness to pay for attributes in preference space (€/year)

Models in WTP space

59In the previous section, we assume that preference heterogeneity is the main reason that individuals make different choices. However, preference heterogeneity can be a consequence of scale heterogeneity. Following recent literature, we move from a preference space framework to a WTP space framework to determine whether our results are robust to the choice of the distribution. Table 5 reports the WTP for each attribute. Model 1’ and model 2’ in the table are analogous to model 1 and model 2 except that they are estimated in WTP space. Both models are estimated using the gmnl command in Stata (Gu, Hole and Knox [2013]). We use 2500 Halton draws for the estimations. As in preference space, the level of fish population is the attribute for which people are most willing to pay (€230 for model 1’ and €220 for model 2’) and hydromorphology the least valued attribute (€110 for model 1’ and €118 for model 2’).

Table 5

Willingness to pay for attributes in WTP space (€/year)

Model 1’Model 2’
Log-likelihood– 1322.8– 1282.1

Willingness to pay for attributes in WTP space (€/year)

60Although it is evident that results from the WTP space approach illustrate the same broad trends as the preference space approach, the mean WTP measures in Table 3 are much lower than those derived from the corresponding models in preference space. Whichever model is considered, the WTP distribution is less skewed than in preference space. Similar to in preference space, significant heterogeneity of preferences is evident in WTP space. This degree of heterogeneity varies depending on the model. For example, for the fish population attributes, the degree of heterogeneity decreases from model 1’ to model 2’ (we observe the opposite effect in preference space).

61Comparing models in the preference space framework and the WTP space framework, it can be observed that models in WTP space do not fit as well as corresponding models in preference space. This result is in line with recent literature (see Train and Weeks [2005]; Sonnier, Ainslie and Otter [2007]; and Hole and Kolstad [2012]). On the other hand, models estimated in WTP space are less likely to return implausible WTP estimates (Owusu Coffie et al. [2016]).

Compensating surplus for different scenarios

62Estimations in Tables 4 and 5 can be used to calculate the total WTP for different management scenarios. Because the utility function that we are using is linear, its value is the sum of its parts. Its attributes can be combined in different ways to estimate welfare effects of discrete changes in the set of attributes.

63This situation can be calculated with the log-sum formula (Hanemann [1999]):


equation im7

65where V1n and V0n represent the utility after and before the change and βp is the marginal utility of money. In our design, we consider the two following scenarios:

  • Scenario 1: From opt out to satisfactory fish population, natural flow and very good water quality;
  • Scenario 2: From opt out to satisfactory fish population, natural flow and sufficient water quality.

66We then derive the compensating surplus for the Aspe households using estimations of both model 2 and model 2’ which allow for unrestricted correlation matrix. Table 4 below reports these results.

Table 6

Compensating surplus (WTP) for different scenarios

Model 2Model 2’
Single HouseholdAspe householdsSingle HouseholdAspe households
Scenario 11,225.0815,926,040922.1411,987,820
Scenario 2905.9611,777,480643.578,366,410

Compensating surplus (WTP) for different scenarios

67As shown in scenario 2, the willingness to pay for a pristine Aspe (that is a satisfactory level of fish population, a very good water quality and a natural flow) is estimated between €922 to €1,225 per household per year. Considering that in the Aspe region there are slightly less than 13,000 households, the cumulative willingness to pay is not far from €12 million, which is three times the revenue sharing that would accrue to local authorities if bidders were to offer 25% revenue sharing. These results unequivocally demonstrate that people living in the region value the fluvial ecosystem.

Restoration costs and ecosystem value

68While the individual WTP can be easily compared with the amount of the rebate, no immediate benchmarking exists for the aggregate results. In order to assess the magnitude of the aggregate WTP, we compare the scenarios discussed above with the restoration costs that were paid by the local authorities to clean up the Aspe river after a major environmental accident, which took place in 2007. A truck discharged 17,000 litres of potassium hydroxide into the river, destroying four kilometers of fauna and severely affecting the upper portion of the river. This led to an immediate ban on fishing that lasted five years and to the adoption of relevant mitigation measures, among which a compulsory release of water by hydropower producers. In particular, hydropower producers had to immediately empty their reservoirs and for subsequent months and years they had to increase the “vital flow,” that is the quantity of water that they had to let into the river. These measures were aimed at diluting pollutants letting hydromorphology mimic a natural state. After seven years, the Aspe ecosystem has completely recovered.

69Direct restoration costs amounted at €121,000, according to the final assessment study carried out by the French Ministry of Environment (MEDDTL [2011]). Unfortunately, no public data are available in terms of foregone profits suffered by hydropower producers; hence, direct restoration costs represent a lower bound. It is important to bear in mind that the operations carried out were only to recover a part of the Aspe to its previous status, the same status that we consider as the status quo in our questionnaire. This means that the results are not directly comparable because they refer to different levels of ecosystem improvements. Still, they can help inform differences between the methods. Considering that approximately 1,200 households live in that same area, we can aggregate our WTP estimates and compare the total to the figures estimated in the restoration cost study.

Table 7

Comparing restoration costs with the compensating surplus for different scenarios

RestorationModel 4Model 6
Cost/Value (€)121,0001,087,152 – 1,470,096772,284 – 1,106,568

Comparing restoration costs with the compensating surplus for different scenarios

70Restoration costs largely underestimate the value of the ecosystem. The cost of restoring the upper part of the Aspe is lower than the perceived value of improvements from the current status to better fluvial ecosystem situations. These results support previous findings that individuals highly value the ecosystem in which they live.


71Hydropower concession renewals are a major challenge for the French electricity market, as hydroelectricity accounts for 24% of total electricity generation. From an economic standpoint, the renewal procedure is an opportunity to redistribute hydropower profits. From an environmental standpoint, the Water Framework Directive requires that all water bodies should attain a good ecological status by 2015 and that water users should pay its full cost, which requires both internalizing environmental externalities and estimating the opportunity cost of the resource. Consequently, the renewal procedure is also an occasion to implement the requirements set forth in the Water Framework Directive.

72The renewal procedure will take the form of a bidding contest, where bidders have to present offers for technical and environmental improvement, as well as a revenue sharing percentage for local authorities. For local communities, in theory, renewal represents a net benefit: either they receive more money or an improved environment. At the same time, however, this framework generates a potential trade-off between revenue-sharing and environmental improvements.

73This paper investigates precisely this trade-off using a discrete choice experiment. Our DCE analysis is unique because we have translated the revenue sharing in an immediate rebate on electricity bills. The local concentration of benefits explains our decision to target households in the Aspe region. The same aspect, that is the geographical boundary of the analysis explains a relatively high estimate of the marginal willingness to pay to increase the ecological health of the Aspe River. The highest total willingness to pay is above €1,225 per household per year. Our results show that people’s marginal willingness to pay for a specific attribute can reach €277 per year, which is three times the maximum rebate that was offered. Moreover, all environmental attributes were considered significant and worth a monetary investment.

74The implication of this study is straightforward: people place considerable value on the improvement of the Aspe ecosystem and this value is higher than actual restoration costs. Therefore, the bidding contest should focus on environmental improvement. Moreover, bidders should react accordingly and offer more environmental improvements. Both potential bidders and authorities should be interested in estimating the value of the fluvial ecosystem and people’s willingness to pay for pristine rivers. This knowledge should inform the structure of the bidding contest and generate bids that are more effective.


The English translation of the survey made in the Aspe Region

75Paris X University and Bocconi University (Italy) are working on a research program, whose purpose is to provide a tool for assessing the environmental costs of operating hydroelectric concessions. The Aspe River is one of the mountain streams that have been selected for this research, which entails a survey to study households’ attitude towards hydropower production.

Aspe River

76The Aspe River is listed as one of the Natura 2000 sites. The Natura 2000 network concerns natural or semi-natural areas of the European Union of great heritage value because of their exceptional flora and fauna.

figure im8
Source: Observatoire de l’eau du bassin de l’Adour.

77We remind you that the survey is anonymous.

Section 1


  1. You are…
    1. Male
    2. Female
  2. Your year of birth
    1. ……………
  3. What is the highest level of education you have completed?
    1. Elementary school
    2. Junior high school
    3. High school
    4. University degree
    5. Other _____________
  4. How many people live in your family (including yourself)?
    1. …………
  5. Your annual income (in euro)?
    1. 0–10,000
    2. 10,001–20,000
    3. 20,001–30,000
    4. 30,001–50,000
    5. 50,001–100,000
    6. Over 100,000
  6. At what distance is the Aspe River from your house?
    1. Less than a kilometer
    2. Between 1 and 5 kilometers
    3. More than 5 kilometers
  7. Do you practice any leisure activity connected to the Aspe?
    1. Fishing
    2. Swimming
    3. Hiking
    4. Rafting
    5. Canoeing
    6. Hunting
    7. Studies and research
    8. Others
    9. No activity
  8. How often do you practice those activities?
    1. Weekly
    2. Monthly
    3. More than once a year
    4. At least once a year
    5. Less than once a year
  9. Are you a member of an environmental organization?
    1. Yes
    2. No
  10. Are you aware of the fact that in the next years hydropower concessions in the Aspe River will expire?
    1. Yes
    2. No

Section 2

79Eight scenarios (choice sets) are presented below. They concern the environmental impacts generated by different ways of managing hydropower.

80We took into account a limited number of environmental attributes and, similarly, we have considered a limited number of levels of variation for each attribute. Although they are not exhaustive, attributes and levels chosen give a precise idea of the ecosystem under study.

81In each scenario, we assume that there are three hydropower producers. Each producer offers annual rebates on your electricity bill. Producer “C” will always offer you the maximum rebate, preserving the current ecosystem status of the Aspe River. On the other hand, producers “A” and “B” will offer smaller discounts, but in each scenario, they will also provide improvements to the Aspe ecosystem.

82For every choice set, you will be asked to choose the producer you prefer. There are no absurd choices.

tableau im9
Choice Set 1 Attributes Producer A Producer B Producer C Fish Sea trout Atlantic salmon Not satisfactory Status and evolution Not satisfactory Status and evolution Not satisfactory Status and evolution Hydromorphology Natural Natural Artificial Physical and chemical water quality Very good Good Sufficient Rebate in euro (on your yearly electricity bill) 10 40 75 Choice
tableau im10
Choice Set 2 Attributes Producer A Producer B Producer C Fish Sea trout Atlantic salmon Not satisfactory Status and evolution Satisfactory Status and evolution Not satisfactory Status and evolution Hydromorphology Natural Natural Artificial Physical and chemical water quality Very good Good Sufficient Rebate in euro (on your yearly electricity bill) 0 10 75 Choice
tableau im11
Choice Set 3 Attributes Producer A Producer B Producer C Fish Sea trout Atlantic salmon Satisfactory Status and evolution Not satisfactory Status and evolution Not satisfactory Status and evolution Hydromorphology Natural Natural Artificial Physical and chemical water quality Sufficient Very good Sufficient Rebate in euro (on your yearly electricity bill) 0 0 75 Choice
tableau im12
Choice Set 4 Attributes Producer A Producer B Producer C Fish Sea trout Atlantic salmon Satisfactory Status and evolution Satisfactory Status and evolution Not satisfactory Status and evolution Hydromorphology Artificial Artificial Artificial Physical and chemical water quality Very good Sufficient Sufficient Rebate in euro (on your yearly electricity bill) 0 10 75 Choice
tableau im13
Choice Set 5 Attributes Producer A Producer B Producer C Fish Sea trout Atlantic salmon Satisfactory Status and evolution Not satisfactory Status and evolution Not satisfactory Status and evolution Hydromorphology Artificial Natural Artificial Physical and chemical water quality Very good Good Sufficient Rebate in euro (on your yearly electricity bill) 40 40 75 Choice
tableau im14
Choice Set 6 Attributes Producer A Producer B Producer C Fish Sea trout Atlantic salmon Satisfactory Status and evolution Satisfactory Status and evolution Not satisfactory Status and evolution Hydromorphology Natural Artificial Artificial Physical and chemical water quality Very good Very good Sufficient Rebate in euro (on your yearly electricity bill) 0 40 75 Choice
tableau im15
Choice Set 7 Attributes Producer A Producer B Producer C Fish Sea trout Atlantic salmon Satisfactory Status and evolution Satisfactory Status and evolution Not satisfactory Status and evolution Hydromorphology Natural Artificial Artificial Physical and chemical water quality Good Very good Sufficient Rebate in euro (on your yearly electricity bill) 10 40 75 Choice
tableau im16
Choice Set 8 Attributes Producer A Producer B Producer C Fish Sea trout Atlantic salmon Not satisfactory Status and evolution Not satisfactory Status and evolution Not satisfactory Status and evolution Hydromorphology Artificial Natural Artificial Physical and chemical water quality Very good Sufficient Sufficient Rebate in euro (on your yearly electricity bill) 10 40 75 Choice
Note: Pictures of the sea trout and the Atlantic salmon are from the Direction Régionale de l’Environnement Aquitaine.


  • [*]
    No specific funding was provided for this research project. The data collected during the survey will be provided upon request.
  • [1]
    In the sample, we did include fishermen as well as neutral citizens. Therefore, in our case it is not possible to disentangle trade-offs specific to each category of users.

In the coming years, a consistent number of French hydropower concessions will be re-assigned through a competitive public procedure, in which bidders will submit an environmental offer and an economic one. This procedure is likely to generate a trade-off between monetary transfers and environmental pledges. In particular, we expect bidders to couple higher monetary compensation to local authorities with lower mitigation measures and vice-versa. This paper aims at investigating the reasons under which local authorities would prefer either environmental protection or monetary transfers. Through an innovative discrete choice experiment (DCE) carried out in the Aspe valley, where more than 100 MW of hydro capacity is installed, we translate monetary offers into immediate rebates on the electricity bill and we test whether local respondents prefer higher rebates or stronger environmental improvements. From a methodological perspective, we calculate our results both with the standard preference space approach and with the WTP space approach. According to our results, locals value considerably more environmental pledges than monetary offers; hence, we argue that bidders should give clear priority to environmental mitigation measures.


  • Water Framework Directive
  • choice experiment
  • hydropower

Implications économiques et environnementales du renouvellement des concessions hydroélectriques : étude de cas dans le sud de la France

Dans les années à venir, les concessions hydroélectriques françaises seront soumises à un processus de mise en concurrence, avec une offre qui prévoit deux volets, économique et environnemental. Cette procédure pourrait générer un arbitrage entre transferts monétaires et engagements environnementaux. En particulier, nous anticipons que des compensations monétaires élevées vis-à-vis des autorités locales seront couplées avec des mesures de mitigation moins satisfaisantes et vice-versa. Cet article analyse les raisons à l’origine des choix des autorités locales auxquelles ces compensations s’adressent. À travers une expérience dans le cadre des modèles de choix discret (DCE) conduite dans la vallée de l’Aspe, où est installée une capacité hydroélectrique de 100 MW, nous considérons que les offres monétaires sont converties en rabais de la facture d’électricité pour les résidents. Nous testons dans ce contexte si la population locale préfère des économies sur la facture plus importantes ou une plus ample amélioration environnementale. Du point de vue méthodologique, nous calculons nos résultats suivant l’approche standard de l’espace des préférences et celle de la propension à payer (willingness to pay, WTP). Nous montrons que la population locale révèle une préférence forte pour les améliorations environnementales. Cela nous emmène à conclure que les acteurs intéressés aux concessions hydroélectriques devraient donner une priorité forte aux mesures de mitigation.
JEL Codes: H23, Q2, Q4, Q5.


  • directive-cadre sur l’eau
  • modèle de choix discret
  • hydroélectricité


  • En ligneAlvarez-Farizo, B., Hanley, N., Barberan, R., and Lazaro, A. [2007]. “Choice Modeling at the ‘Market Stall’: Individual versus Collective Interest in Environmental Valuation,” Ecological Economics, 60 (4): 743–751.
  • En ligneBalcombe, K., Bailey, A., Chalak, A. and Fraser, I. [2008]. “Modifying Willingness to Pay Estimates Where Respondents Misreport their Preferences,” Applied Economic Letter, 15 (5): 327–330.
  • En ligneBalcombe, K., Chalak, A. and Fraser, I. [2009]. “Model Selection for the Mixed Logit with Bayesian Estimation,” Journal of Environmental Economics and Management, 57 (2): 226–237.
  • Bateman, I., Carson, R., Day, B., Hanemann, M., Hanley, N., Hett, T., Jone-Lee, M., Loomes, G., Mourato, S., Ozdemiroglu, E., Pearce, D., Sugden, R. and Swanson, J. [2002]. Economic Valuation with Stated Preference Techniques: A Manual. Cheltenham: Edward Elgar.
  • Birol, E., Koundouri, P. and Kountouris, Y. [2008]. “Using the Choice Experiment Method to Inform River Management in Poland: Flood Risk Reduction vs. Habitat Conservation in the Upper Silesia Region.” In Birol, E. and Koundouri, P. (eds). Choice Experiments Informing Environmental Policy: A European Perspective. Cheltenham: Edward Elgar, p. 365–394.
  • En ligneBrouwer, R., Martin-Ortega, J. and Berbel, J. [2010]. “Spatial Preference Heterogeneity: A Choice Experiment,” Land Economics, 86 (3): 552–568.
  • En ligneBrown, L. E., Hannah, D. M. and Milner, A. M. [2007]. “Vulnerability of Alpine Stream Biodiversity to Shrinking Glaciers and Snow-Packs,” Global Change Biology, 13: 958–966.
  • En ligneBurton, M. and Rigby, D. [2009]. “Hurdle and Latent Class Approaches to Serial Non-Participation in Choice Models,” Environmental and Resource Economics, 42 (2): 211–226.
  • En ligneCarlsson, F., Frykblom, P. and Lagerkvist, C. J. [2005]. “Using Cheap Talk as a Test of Validity in Choice Experiments,” Economic Letters, 89 (2): 147–152.
  • En ligneCéréghino, R., Cugny, P. and Lavandier, P. [2002]. “Influence of Intermittent Hydro-Peaking on the Longitudinal Zonation Patterns of Benthic Invertebrates in a Mountain Stream,” International Review of Hydrobiology, 87: 47–60.
  • En ligneChanseau M., Larinier, M. and Travade, F. [1999]. “Efficacité d’un exutoire de dévalaison pour smolts de saumon atlantique (Salmo salar L.) et comportement des poissons au niveau de l’aménagement hydroélectrique de Bedous sur le gave d’Aspe étudiés par la technique de marquage-recapture et par radiotélémétrie,” Bulletin France Pêche Piscicole, 353–354: 99–120.
  • COGEPOMI [2008]. Plan de gestion de poissons migrateurs, 2008–2012: Adour et cours d’eau côtiers. Bordeaux: Direction régionale de l’Environnement Aquitaine.
  • Commission de Régulation de l’Énergie [2013]. Activity Report 2012. Paris: Commission de Régulation de l’Énergie.
  • Compagnie Nationale du Rhône [2013]. Rapport d’activité 2012. Lyon, Compagnie Nationale du Rhône.
  • Crastes, R. and Mahieu, P.-A. [2014]. “Contingent Valuation versus Choice Experiments: A Meta-Analysis Application Exploring the Determinants of the Time for Publication Acceptance,” Economics Bulletin, 34 (3): 1575–1599.
  • En ligneCroze, O., Bau, F. and Delmouly, L. [2008]. “Efficiency of a Fish Lift for Returning Atlantic Salmon at a Large-Scale Hydroelectric Complex in France,” Fisheries Management and Ecology, 15: 467–476.
  • En ligneCummings, R. G. and Taylor, L. [1999]. “Unbiased Value Estimates for Environmental Goods: A Cheap Talk Design for Contigent Valuation Method,” The American Economic Review, 89 (3): 649–665.
  • En ligneDoherty, E., Campbell, D. and Hynes, S. [2013]. “Models of Site-Choice for Walks in Ireland: Exploring Cost Heterogeneity,” Journal of Agricultural Economics, 64 (2): 446–466.
  • En ligneFiebig, D. G., Keane, M. P., Louviere, J. and Wasi, N. [2010]. “The Generalized Multinomial Logit Model: Accounting for Scale and Coefficient Heterogeneity,” Marketing Science, 29 (3): 393–421.
  • En ligneFosgerau, M. and Bierlaire, M. [2007]. “A Practical Test for the Choice of Mixing Distribution in Discrete Choice Models,” Transportation Research Part B: Methodological, 41 (7): 784–794.
  • En ligneGu, Y., Hole, A. R. and Knox, S. [2013]. “Fitting the Generalized Multinomial Logit Model in Stata,” The Stata Journal, 13 (2): 382–397.
  • En ligneGreene, W. and Hensher, D. [2010]. “Does Scale Heterogeneity across Individuals Matter? An Empirical Assessment of Alternative Logit Models,” Transportation, 37 (3): 413–428.
  • En ligneHalkos, G. and Matsiori, S. [2014]. “Exploring Social Attitude and Willingness to Pay for Water Resources Conservation,” Journal of Behavioral and Experimental Economics, 49: 54–62.
  • Hanemann, M. [1999]. “Welfare Analysis with Discrete Choice Models.” In Herriges, J. and Kling, C. (eds). Valuing Recreation and the Environment. Cheltenham: Edward Elgar, p. 33–64.
  • En ligneHanley, N., Adamowicz, W. and Wright, R. [2005]. “Price Vector Effects in Choice Experiments and Empirical Test,” Resource and Energy Economics, 27: 227–234.
  • En ligneHanley, N., Wright, R. and Alvarez-Farizo, B. [2006]. “Estimating the Economic Value of Improvements in River Ecology Using Choice Experiments: An Application to the Water Framework Directive,” Journal of Environmental Management, 78 (2): 183–193.
  • En ligneHicks, J. R. [1943]. “The Four Consumer’s Surpluses,” The Review of Economic Studies, 11 (1): 31–41.
  • En ligneHole, A. R. [2007]. “Fitting Mixed Logit Models by Using Maximum Simulated Likelihood,” The Stata Journal, 7 (3): 388–401.
  • En ligneHole, A. R. and Kolstad, J. R. [2012]. “Mixed Logit Estimation of Willingness to Pay Distributions: A Comparison of Models in Preference and WTP Space Using Data from a Health-Related Choice Experiment”, Empirical Economics, 42 (2): 445–469.
  • En ligneHoyos, D. [2010]. “The State of the Art of Environmental Valuation with Discrete Choice Experiments”, Ecological Economics, 69 (8): 1595–1603.
  • En ligneHynes, S., Hanley, N. and Scarpa, R. [2008]. “Effects on Welfare Measures of Alternative Means of Accounting for Preference Heterogeneity in Recreational Demand Models,” American Journal of Agricultural Economics, 90 (4): 1011–1027.
  • INSEE [2013]. Dossier complet. Département des Pyrénées-Atlantiques (64),
  • International Renewable Energy Agency [2012]. “Hydropower,” IRENA Working Paper, 3/5.
  • En ligneKataria, M. [2009]. “Willingness to Pay for Environmental Improvements in Hydropower Regulated Rivers,” Energy Economics, 31 (1): 69–76.
  • En ligneKataria, M., Bateman, I., Christensen, T., Dubgaard, A., Hasler, B., Hime, S., Ladenburg, J., Levin, G., Martinsen, L. and Nissen, C. [2012]. “Scenario Realism and Welfare Estimate in Choice Experiments: A Non-Market Valuation Study on the European Water Framework Directive,” Journal of Environmental Management, 94 (1): 25–33.
  • En ligneKragt, M., Newham, L., Bennett, J. and Jakeman, A. [2011]. “An Integrated Approach to Linking Economic Valuation and Catchment Modelling,” Environmental Modelling & Software, 26 (1): 92–102.
  • Louviere, J., Hensher, D. A. and Swait, J. [2000]. Stated Choice Methods: Analysis and Applications. Cambridge: Cambridge University Press.
  • McFadden, D. [1973]. “Conditional Logit Analysis of Qualitative Choice Behavior.” In Zarembka, P. (ed.). Frontiers in Econometrics. New York: Academic Press, p. 105–142.
  • MEDDTL [2011]. Application des méthodes d’équivalence à la pollution accidentelle du gave d’Aspe. Paris: Ministère de l’Écologie, du Developpement Durable, de Transport et du Logement.
  • En ligneMetcalfe, P., Baker, W., Andrews, K., Atkinson, G., Bateman, I., Butler, S., Carson, R., East, J., Gueron, Y., Sheldon, R. and Train, K. [2012]. “An Assessment of the Nonmarket Benefits of the Water Framework Directive for Households in England and Wales,” Water Resource Research, 48 (3): 18–31.
  • Mitchell, R. and Carson, R. [1989]. Using Surveys to Value Public Goods: The Contingent Valuation Method. Washington (D. C.): Resources for the Future Press.
  • En ligneMorrison, M. D. and Bennett, J. [2004]. “Valuing New South Wales Rivers for Use in Benefit Transfer,” Australian Journal of Agricultural & Resource Economics, 48 (4): 591–611.
  • En ligneMorrison, M. D., Blamey, R. K. and Bennett, J. W. [2000]. “Minimising Payment Vehicle Bias in Contingent Valuation Studies,” Environmental and Resource Economics, 16 (4): 407–422.
  • En ligneOwusu Coffie, R., Burton, M. P., Gibson, F. L. and Hailu, A. [2016]. “Choice of Rice Production Practices in Ghana: A Comparison of Willingness to Pay and Preference Space Estimates,” Journal of Agricultural Economics, 67 (3): 799–819.
  • Pearce, D., Atkinson, G. and Mourato, S. [2006]. Cost-Benefit Analysis and the Environment: Recent Developments. Paris: Organisation for Economic Co-operation and Development.
  • Poirier, J. and Fleuret, A. [2010]. “Using the Choice Experiment Method for Valuing Improvements in Water Quality: A Simultaneous Application to Four Recreation Sites of a River Basin,” paper presented at the 59th Annual Meeting of the French Economic Association (AFSE), Université Paris-Nanterre.
  • En ligneRenofalt, B. M., Jansson, R. and Nilsson, C. [2010]. “Effects of Hydropower Generation and Opportunities for Environmental Flow Management in Swedish Riverine Ecosystems,” Freshwater Biology, 55 (1): 49–67.
  • En ligneScarpa, R., Thiene, M. and Train, K. [2008]. “Utility in Willingness to Pay Space: A Tool to Address Confounding Random Scale Effects in Destination Choice to the Alps,” American Journal Agricultural Economics, 90 (4): 994–1010.
  • En ligneSonnier, G., Ainslie, A. and Otter, T. [2007]. “Heterogeneity Distributions of Willingness-to-Pay in Choice Models,” Quantitative Marketing and Economics, 5 (3): 313–331.
  • Stithou, M., Hynes, S., Hanley, N. and Campbell, D. [2012]. “Estimating the Value of Achieving ‘Good Ecological Status’ in the Boyne River Catchment in Ireland Using Choice Experiments”, The Economic and Social Review, 43 (3): 397–422.
  • En ligneTrain, K. B. [2003]. Discrete Choice Methods with Simulation. Cambridge: Cambridge University Press.
  • Train, K. B. [2009]. Discrete Choice Methods with Simulation. Cambridge: Cambridge University Press, 2nd ed.
  • En ligneTrain, K. B. and Weeks, M. [2005]. “Discrete Choice Models in Preference Space and Willingness-to-Pay Space.” In Scarpa, R. and Alberini, A. (eds). Application of Simulation Methods in Environmental and Resource Economics. Dordrecht: Springer, p. 1–16.
Federico Pontoni
IEFE. Correspondence: Università Bocconi, via Rontgen 1, 20136, Milan, Italy.
Anna Creti
Université Paris-Dauphine, PSL Research University, LEDa. Correspondence: Place du Maréchal-de-Lattre-de-Tassigny, 75016 Paris, France.
Marc Joëts
Banque de France, International Macroeconomic Division, and EconomiX CNRS, University of Paris Ouest, France. Correspondence: Banque de France, International Macroeconomic Division, DGEI-DERIE-SEMSI, 39 rue Croix-des-Petits-Champs, 75049 Paris Cedex 01, France.
Cette publication est la plus récente de l'auteur sur
Dernière publication diffusée sur ou sur un portail partenaire
Cette publication est la plus récente de l'auteur sur
Mis en ligne sur le 16/03/2018
Pour citer cet article
Distribution électronique pour Presses de Sciences Po © Presses de Sciences Po. Tous droits réservés pour tous pays. Il est interdit, sauf accord préalable et écrit de l’éditeur, de reproduire (notamment par photocopie) partiellement ou totalement le présent article, de le stocker dans une banque de données ou de le communiquer au public sous quelque forme et de quelque manière que ce soit.
Chargement en cours.
Veuillez patienter...