Impact of justice and solidarity variables on the acceptability of managed realignment

Sea-level rise due to climate change will have significant effects on coastal areas and populations. Adaptation policies recommend the managed realignment of the most vulnerable assets and activities. Despite their medium- and long-term benefits, these policies face significant friction due to social acceptability in the communities where they are implemented. This article investigates the hypothesis that respecting principles of justice in the implementation of managed realignment should increase its acceptability. We compare preferences of those people who are exposed to the risk of climate-change-induced flooding and those who are not, as regards funding managed retreat policies and defining compensation criteria for assets at risk. The main theories of social justice provide the four principles included in the analysis: efficiency, need, responsibility and priority assigned to property rights. A choice experiment survey was conducted with 258 residents of coastal and hinterland communities in the south of France. Four attributes were selected to define the managed realignment policy: the dialogue arrangements, the implementation period, the policy implementation schedule and the cost. The results show support for a relatively fast launch of these policies (within 15 years) but in stages and through a process of dialogue with the population. People's perceptions of the funding criteria reveal a preference for national solidarity. Finally, national funding of managed retreat policies and compensation criteria based on market prices have a significant positive influence on the acceptability of managed realignment policies, whereas introducing responsibility-based compensation criteria tends to favour the status quo over the adaptation policy. Policy relevance Prioritization of the funding criteria reveals the preference for national solidarity. Preferences for the justice criteria underpinning compensation reveal a great diversity of values. Besides implantation modalities, preferences for managed realignment policies depend on which level they are implemented at, on the expropriation criteria (the emphasis given to property rights, i.e. market price), on the attachment (people perceived as worst off, i.e. the property is their main residence rather than a second home or they have lower levels of income) and on the degree of responsibility (related to the date of purchase, i.e. on the information given at the time on the risk).


Introduction
The issue of climate change adaptation is particularly relevant where the vulnerability of coastal areas is concerned. Sea-level rise will result in the permanent inundation of some areas and increased storm damage due to the higher water levels. In the case of Languedoc-Roussillon (France), this rise has been estimated to be 40 cm by 2060 based on an adaptation of the Intergovernmental Panel on Climate Change (IPCC) hypotheses (Magnier, 2013). At that date, water levels currently reached during 100-year events may have a decennial return period. Faced with this outlook, shoreline management policies require adaptation. It is recommended that vulnerability is reduced by relocating shoreline assets and activities to retreat areas further inland (MEDDE, 2012). Such policies are of interest not only because of the future savings in damage costs but also because they protect dune biodiversity whilst maintaining the recreational services offered by beaches (Cooper & Lemckert, 2012).
However, the social acceptability and the economic feasibility of these managed realignment policies, even when restricted to a few neighbourhoods, are not straightforward. Besides the uncertainty, and even the scepticism, towards the evolution of these long-term risks that characterize any climate change adaptation policy, managed realignment is hindered by residents' attachment to their assets and the amenities provided by the proximity to the sea. An extensive literature focuses on the dimensions and predictors of the social acceptability of adaptation policies and on the notion of 'social acceptance' in relation, for instance, to 'social support' (Batel, Devine-Wright, & Tangeland, 2013). Wüstenhagen, Wolsink, and Bürer (2007) identify three types of acceptability that are often interactive, depending on the scale: (1) socio-political acceptability where stakeholders are concerned, (2) community acceptability where residents are concerned, and (3) market acceptability where consumers and investors are concerned. As regards acceptability factors, previous work has shown that day-trippers and tourists are significantly more in favour of managed realignment policies than local residents (Rey-Valette et al., 2012;Rulleau & Rey-Valette, 2013), particularly second-home owners who are often older and more attached to their assets (Rey-Valette, . Opposition is all the stronger in that, depending on the area's configuration, retreat areas may be a significant distance away, even in other towns or villages. Such relocations require coherent territorial planning at a large scale (Hurlimann et al., 2014) and the mobilization of solidarity between communities. In an institutional approach, solidarity can be defined using social capital parameters that play a decisive role in acceptability. As Jones and Clark (2014) show, the level of social trust within the coastal communities has a significant impact on residents' perceptions of the costs and benefits of adaptation policies. In this article, solidarity and justice principles are defined on the basis of the main theories of social justice. The hypothesis under investigation is that respecting these principles in the implementation of managed realignment increases its acceptability. In funding managed retreat policies, solidarity mostly concerns those people who are exposed to the risk of climate-change-induced flooding and those who are not. In addressing damage compensation mechanisms, four justice principles were selected for our analysis: efficiency, need, responsibility and priority assigned to property rights.
More specifically, the objective was to examine whether people's perceptions of solidarity and justice in realignment policies affect individual preferences for the managed realignment policy compared with the laissez-faire alternative. To this end, a choice experiment (CE) was used to elicit individual preferences regarding the characteristics of managed realignment policies. This economic valuation method is grounded in Lancaster's characteristics theory (1966), which hypothesises that the satisfaction obtained by consumers from particular goods comes in fact from the consumption of the different characteristics that these goods embody. It is used specifically to study consumers' trade-offs between the different components of a particular good (Bennett & Blamey, 2001). We use it to estimate the value that people place on the attributes of a managed realignment policy, focusing on the hypothetical trade-offs that they make between attributes and the methods used to implement the policy. Thus, our study should provide more information than contingent valuation to the managers and decision-makers involved in the design of acceptable adaptation policies (Hoyos, 2010).
CE is applied in many areas (Hanley & Barbier, 2009). In the case of climate change, most studies focus on individuals' preferences for adaptation and/or mitigation policies (Akter, Bennett, & Ward, 2012;Carson, Louviere, & Wei, 2010;Rajmis, Barkmann, & Marggraf, 2009), or ecosystem protection (Koundouri et al., 2012;Riera, Peñuelas, Farreras, & Estiarte, 2007;Shoyama, Managi, & Yamagata, 2013), on individuals' adaptive capacity (Alberini, Chiabai, & Muehlenbachs, 2006;Brouwer & Schaafsma, 2013) or on climate change impacts on insurance (Brouwer & Akter, 2010;Wouter Botzen & Van Den Bergh, 2012). There are few studies applied, like ours, to the case of coastal zones. Remoundou, Diaz-Simal, Koundouri, and Rulleau (2015) carried out a CE in Spain where the attributes used related to the loss of beach size and associated recreational aspects, the change in biodiversity and the health risks due to the presence of jellyfish. More in keeping with our issue, Luisetti et al. (2011) examined managed realignment in England. But they focused on the site's recreational services whereas we studied the precise methods to implement the managed realignment policy. Zander, Petheram, and Garnett (2013) also addressed relocation but their study focused on Australian aborigines. The article contributes to this limited literature using the example of the French coastal zone.
In section 2, the acceptability constraints of the managed realignment policies are analysed. Section 3 sets out the methodology and the characteristics of the survey protocol. The main results are then presented in Section 4 and discussed in the Section 5 part. Section 6 provides concluding comments.

The acceptability of managed realignment policies
Numerous factors hinder the acceptability of managed realignment policies. First, the fact that the shoreline moves naturally must be accepted. Climate change adaptation requires the recognition of the natural resilience limits of non-urbanized beaches. The issue is to reduce coastal area vulnerability by relocating the assets that are most at risk (Clark, 1998;Kelly & Adger, 2000;Titus, 1998) rather than trying systematically to build sea defences, which are now known to worsen erosion (Pilkey, 1990). This kind of profound change in representation implies a learning process and specific communication methods, especially as the progressive nature of the risks related to climate change and the numerous associated uncertainties strengthen optimism and status quo biases (Lammel, Dugas, & Guillen Gutierrez, 2012;Lammel, 2013). Spontaneous behaviour tends towards a 'wait-and-see' attitude (Dutt & Gonzales, 2012) and depends heavily on perceptions. These, in turn, depend on socio-demographic characteristics such as the age, education and status of main residents or second-home owners (Hellequin, Flanquart, Meur-Férec, & Rulleau, 2014;Rey-Valette et al., 2012. In addition to this change in reference framework, other factors complicate the implementation of managed realignment policies. Several studies have shown the importance of dialogue and of the legitimacy of the institution in charge of implementation (Cooper & Pile, 2014;Kelly & Adger, 2000;King et al., 2014;Ledoux, Cornell, O'Riordan, Harvey, & Banyard, 2005;Myatt, Scrimshaw, & Lester, 2003;Roca & Villares, 2012). Acceptability must be considered with respect to elected representatives and the local population. Elected representatives and managers are concerned about the scope of territorial reorganization and the financial requirements for compensation (French, 2006;King et al., 2014;Ledoux et al., 2005), and first and foremost by possible opposition from the population. Yet, residents hold contrasting viewpoints depending on their demographic and geographical situation, which evolve when climatic events occur. Several surveys in Australia have shown that the desire to move increases (from 30% to 36%) following a severe storm and, on the other hand, that opposition increases in case of a significant drop in the housing market, especially with indebted homeowners (King et al., 2014).
More generally, whilst hinterland residents tend to favour managed realignment policies as they wish to maintain the beaches' recreational amenities, residents who are directly concerned are usually opposed to these policies as they are attached to their assets and the amenities provided by their proximity to the sea. However, compensation levels together with expropriation and buy-back procedures and conditions are key elements for acceptability (French, 2006;Glenk & Fisher, 2010). These studies have all shown the need to propose a coherent buy-back policy and the importance of governance conditions and dialogue (Milligan, O'Riordan, Nicholson, & Watkinson, 2009;Penning-Rowsell et al., 2014), which promote trust in institutions (Gibbs, Thebaud, & Lorezn, 2013;Graham et al., 2013;Midgley & McGlashan, 2004;Storbjork & Hedren, 2011).
Such social resistance leads many researchers to stress the need to address better the social determinants and impacts (Adger et al., 2009;Graham et al., 2013;Wise et al., 2014). Several authors suggest that the analysis of justice and equity principles, insufficiently developed at this level 1 (Cooper & McKenna, 2008) would promote managed realignment acceptability (Cooper & Lemckert, 2012;Duxbury & Dickinson, 2007;Glenk & Fisher, 2010;Milligan et al., 2009). Indeed, according to Field et al. (2012), fairer policies are better accepted and are therefore more efficient. Glenk and Fisher (2010) showed the importance and the pluralism of the values mobilized (sustainability, efficiency, solidarity). Likewise, Abel et al. (2011) advocate greater transparency in public or private cost and benefit distribution as well as taking responsibility into account. For example, one issue is to avoid tax payers funding the well-being of the wealthiest, as decried by Pilkey (1990) and Platt, Salvesen, and Baldwin (2002), and another is to review the concept of land property, as recommended by Buhot et al. (2012). These authors, for example, recall the intense controversy over private property that opposed the 'libertarian' and 'liberal-egalitarian' currents, the former initiated by Nozick (1974) and the latter associated with Rawls (1971).
In line with this debate, we investigated the justice principles that might underpin the methods to implement managed realignment policies and hence promote their acceptability. To that effect, we used a CE with attributes corresponding to the implementation procedure and then studied the significance of the justice and solidarity variables for relocation policy preferences.

Description of the study zone
The survey zone was the SCoT 2 territory around Béziers, in Languedoc-Roussillon (southeastern France), where the coastal towns are among the most vulnerable in the area to marine erosion and inundation (MIAL-LR, 2003). We focused on two coastal towns (Valras-Plage and Vendres) and three towns at an increasing distance from the coast: Béziers, the economic capital situated 15 km from the sea, Murviellès-Béziers and Saint Chinian, two agricultural hinterland towns 29 and 40 km from the sea, respectively. These towns and villages were selected due to their potential economic and institutional solidarity with their coastal counterparts. They all belong to the same territorial planning institutional system (the Territorial Coherence Scheme, SCoT), which establishes institutional solidarity. Our hypothesis was that solidarity decreases as one moves from the coast towards the hinterland. Solidarity depends on beach uses, on perceptions and on people's attachment to the coast as well as on economic, institutional and ecological interactions between coastal and non-coastal towns and villages (see supplementary data, Figure S1).

Questionnaire, completion method and sample
The questionnaire comprised six parts. 3 The first three parts focused on housing, attachment to the coast, beach usage, coastal knowledge and inundation risk perception. The fourth part comprised the choice sets of the CE and follow-up questions. It was introduced by a presentation of the reference situation and the attributes. The next part related to solidarity and justice issues, whilst the final one focused on the socio-demographic characteristics. Two different versions of the first three parts were designed depending on whether the respondent lived on the coast or in the hinterland. The last three were the same for all respondents. This questionnaire was tested with 30 people in April 2013.
Given our hypotheses, a stratified sampling design was adopted. Level 1 consisted of towns and villages that were selected to cover the diversity of the situations (see supplementary data, Table S1). On the coast, Vendres depends both on tourism and wine production whereas Valras-Plage is the archetype coastal resort with a very large number of second homes. In the hinterland, Béziers is the largest town in the area. Further inland, Murviel les Béziers is a typical suburban wine-producing town whilst Saint Chinian is a typical wine-producing village with rural tourism appeal. In level 2, the survey population consisted of residents who had their main home in the selected towns and villages. Participants were contacted by telephone; they were selected randomly, balancing the numbers from the different towns and villages. The telephone call referred only to coastal planning to avoid introducing bias. The questionnaire was completed in rooms provided by the town councils during May and June 2013. It was self-administered in the presence of 7 researchers available to answer any queries. In total, 13 sessions took place involving, on average, 20 people per session. Sessions alternated between week-days and Saturdays to accommodate working people. The survey took roughly an hour.
In all, 258 people were interviewed: 122 from the coast and 136 from the hinterland. Almost half (48%) of the respondents were native to the territorial 'department' and 54% were women. Just over a third of them (38%) were single and households had an average size (2.4 people) equal to the average in the department. The average age was quite high (53 years), which is common in coastal and rural towns and villages. However, educational and income levels were higher than the departmental average. Almost half of the respondents (46%) had higher-education degrees against 27% for the departmental average (data from the French National Institute for Statistics and Economic Studies for 2012), whilst 22% had an average monthly income lower than the average wage and 27% had an average monthly income of at least EUE3,000. These characteristics may be explained by the fact that a high proportion of property owners was deliberately selected in the sample (73% against 55% on average in the department).
The data were processed using Sphinx to test correlations (chi-square tests) and a mixed logit model was then developed using Nlogit to examine the choices formulated in the CE.

Proposed scenarios of the CE
The attributes and levels used are presented in Table 1. The reference situation was defined according to the IPCC regional forecast. By 2060, without further public intervention, beaches are expected progressively to lose a minimum 30% of their size with very strong 5-to 6-yearly storms, leading to the introduction of insurance premiums proportional to the risk incurred. Faced with this outlook, asset and activity realignment policies may be implemented according to different modalities that were used as attributes in our CE.
Four attributes were selected based on the results of Gordon, Chapman, and Blamey (2001) and the recommendations of Bateman et al. (2002). They were defined on the basis of expert opinion by researchers and State services in charge of these policies ('Direction des risques de la Direction Régionale de l'Environnement, de l'Aménagement et du Logement': DREAL).
Each attribute was described by three levels that were sufficiently different to reflect all future possible options (Bennett & Blamey, 2001). The levels of the monetary attribute were calculated using the cost of the work (demolition and sand nourishment of the areas) estimated by the risk service of DREAL without taking into account owners' compensation. We assumed that half of the work would be funded (2) Coastal infrastructure and houses or shops on the shoreline for a cost of E100 (3) Coastal infrastructure and houses or shops on and behind the shoreline for a cost of E200 by subsidies and the remainder shared between fiscal households at the 'intercommunal' level (cooperation between several towns), the relevant scale for this type of operation. An increase in the local council tax was proposed as the means of payment. We then combined these attributes into alternatives and the alternatives into choice sets. To do this, we used a fractional factorial design and the Ngene software. Based on a method proposed by Zwerina, Huber, and Kuhfeld (2004), this approach ensures that the choice sets best meet the efficiency criteria defined by Huber and Zwerina (1996). The 18 choice sets identified were then divided into three blocks using the method proposed by Louviere (1988) so that each respondent is only presented with six choice sets (Adamowicz, Louviere, & Swait, 1998). The three versions of the questionnaire that correspond to the three blocks of choice sets were allocated randomly and uniformly to the respondents.
Each choice set included the reference situation, which described the current situation with no tax increase (business-as-usual), and two alternatives, corresponding to certain levels of the managed realignment attributes, and for each scenario the financial contribution depending on the size of the area concerned by managed realignment.

Solidarity and justice questions
Given our hypothesis that the respect of justice criteria in the implementation of managed realignment measures enhances their acceptability, our primary aim was to identify, for some factors, the implementation methods that were considered to be the fairest or the most equitable and which, as such, can be considered to be legitimate. In the survey, respondents were asked to rank the principles that they considered to be the fairest regarding both the funding of the managed retreat policy and the damage compensation mechanism.
An ex ante perspective was taken of the financing of retreat policies, highlighting the solidarity between those faced with risk and those not. The funding methods indicate whether the respondents thought that adaptations should benefit from financial solidarity or not, and if so at what level. Solidaristic funding mechanisms proposed in the questionnaire included the choice between a national or a regional tax levy. Respondents could also choose a non-solidaristic funding method such as individual insurance policies, a tourist tax or local taxes in the coastal towns. In the latter case, the town concerned was assumed to be solely responsible for funding the climate risk.
Moreover, managed realignment policies may imply the purchase or expropriation of shoreline assets. The question then arises as to the criteria that should underpin compensation mechanisms for the expropriated assets. It was to address this question that theories of justice were used. The choice of compensation criteria arises in an ex post perspective that highlights fairness between individuals affected by retreat policies. Four general fairness principles were considered relating to the main theories of social justice: (i) efficiency in the sense of the utilitarian principle of maximizing the total surplus (Harsanyi, 1955), (ii) need in the sense of priority to the worst-off defined by Rawls (1971), (iii) responsibility following Dworkin's (1981) Luck Egalitarianism, and finally (iv) priority to property rights in line with the libertarian approach of Nozick (1974). Placing these four principles in the context of relocalization polices enabled compensation criteria to be defined that modulate compensation for expropriation (Clément, Rey-Valette, & Rulleau, 2015). Hence, compensation for expropriated assets at their market value represents an approach based on efficiency and respect for property rights; taking into account individuals' income, and prioritizing main residences over second homes represents an approach based on need; and finally varying compensation according to whether or not individuals were aware of the risk when choosing where to reside integrates the principle of responsibility. We tested these criteria ( Table 2) asking respondents what would be the fairest compensation approach if houses and shops near the beaches were to be expropriated.
Finally, the possible involvement of respondents in solidarity mechanisms was identified. Some 55% of the respondents belonged to one or more associations, but only 19% of the associations dealt with the social sector. Finally, 20% of the respondents declared that they had made regular donations towards solidarity operations or actions in the previous 12 months, compared with 45% who had done so occasionally and 35% who never had.

Willingness to pay modelling
The determinant variables for the econometric modelling are described in Table 3 and the mixed logit results in Table 4. We found that all levels of the attributes had a statistically significant coefficient. In other words, the characteristics of the policies presented to respondents do affect the probability of choosing an alternative policy to the status quo, i.e. a realignment policy, as in our case the status quo implies laissez faire, which means not anticipating the effects of sea-level rise. Both the cost of the realignment policy and any delay in its implementation have a negative effect on this probability whilst dialogue and implementation in several stages have a positive impact.
The other variables were exclusively qualitative, relating to the respondents' characteristics or their perceptions. Several variable categories have an impact on preferences, beginning with those concerning locality and socio-demographics. Living on the coast has a positive effect on the choice of an alternative other than the reference situation. With respect to socio-demographic factors, education and income play a part but, paradoxically, with negative coefficients. Other than a correlation with location in favour of coastal towns and villages where average income is higher, this result may be explained by the stronger attachment of middle and lower classes to the recreational services provided Market value of assets 29 Efficiency and property rights (Harsanyi, 1955;Nozick, 1974) Date of purchase: in favour of longer-term owners 10 Type of housing: in favour of main residence 12 Need (Rawls, 1971) Owner's income: in favour of lower incomes 30 Date of purchase: at the expense of informed owners 14 Responsibility (Dworkin, 1981) Other or no answer 5 -Total 100 -Source: 2013 Solter survey.
by the coast (Rulleau, 2008). Finally, 'intercommunalité' (a group of and cooperation between neighbouring towns and villages) appears to be the most coherent scale for territorial planning to address climate change adaptation, in keeping with the results of other studies (Abel et al., 2011;Graham et al., 2013;Hurlimann et al., 2014). The willingness to pay (WTP) of each individual for a level of attribute can be calculated as follows (Hanemann, 1984): where b c is the coefficient of one of the attribute levels and b y is the coefficient of the monetary attribute. Their variance was calculated using the delta method (Hole, 2007). The values obtained (Table 5) are the total WTP by household. To be in total disagreement with the statement: "compensation at book value rather than market value is unfair as it places at a disadvantage people who have owned their assets for a long time and are therefore more attached to them".

Responsibility
To be in total disagreement with the statement: "it is unfair to use public funding to compensate residents who have recently settled in the area as these people were duly informed of the risk when they made their decision".
The WTP are significant for all levels of attributes but indicate a hierarchy in the implementation arrangements. The respondents ranked dialogue very high on their list as regards the choice of the zones to be relocated. These findings are consistent with the literature on the acceptability of managed realignment policies. The respondents were also willing to pay for this managed realignment to be organized in several stages rather than on a one-off basis. But the willingness to pay for this policy to be delayed was negative. In other words, the respondents preferred the policy to be implemented relatively quickly. In short, the respondents' preferred scenario comprised a managed realignment policy implemented in the next 15 years, in successive stages and through dialogue with the population.

Variables Coefficients
Attributes Log likelihood 2 1,120.50 McFadden pseudo R 2 0.27 ***significant at the 1% level. **significant at the 5% level. *significant at the 10% level. Standard-errors are in brackets. Note: NsCost stands for the derived standard deviation of parameter distribution (normal distribution), the parameter being here the cost.
Among the individual characteristics affecting WTP, and as observed in other surveys (Glenk & Fisher, 2010;Hellequin et al., 2014;King et al., 2014;Myatt et al., 2003;Rey-Valette et al., 2012;Rulleau & Rey-Valette, 2013), we note the impact of education and income, whilst age has no influence. Yet age may indicate a greater degree of reluctance from older people . Surveys undertaken in Australia confirm the lower mobility of people with low incomes and who are older. The authors link age both with the feeling of belonging to a community and the fact that older people feel less concerned with long-term risks (King et al., 2014). The fact that age has no significant impact in our case could be explained by the relatively large number of older people in our sample.
Moreover, we observe that those people most likely to be affected by inundation are more favourable to managed realignment that would benefit them directly. It should be noted that only 30% of coastal respondents lived in areas exposed to inundation that might lead to a managed realignment programme. By enabling beaches and the attractiveness of coastal towns and villages to be maintained, in particular for tourists, managed realignment is of obvious interest for residents who will not have to move. Hinterland residents have a similar interest, although to a lesser extent, as most of them use the beaches (80% of them go to the beach over 10 times per annum). It seems logical, therefore, that respondents whose choice of home was influenced by proximity to the coast, whether they live on the coast or in the hinterland, are less likely to oppose managed realignment policies.

Discussion
The aim of this article was to investigate whether solidarity and justice variables affected individual preferences concerning managed realignment. Three main results emerge from the analysis.
The first result relates to solidarity: respondents preferred managed realignment policies funded on the basis of national solidarity and this variable significantly and positively affected preferences in favour of a realignment policy (Table 4). In the funding of a strategic retreat, the results showed that 47% of interviewees preferred financial solidarity at the national level and 16% at a regional level whilst 37% were opposed to solidarity between those who were faced with a threat and those who were not. In this latter case, funding would come only from individuals or towns directly concerned with 18% choosing tourist taxes, 12% private insurance and 7% an increase in local taxes in the towns concerned.
It should be emphasized that the respondents' support for national solidarity in funding realignment policies is fully in line with current French practice, where beaches are perceived as public goods that are mainly the responsibility of the State, the historical manager of the public maritime domain in France. Hence, in France, damages due to natural disasters and shoreline management benefit from national solidarity. Damages are funded by a major natural risk prevention fund, called the 'Barnier fund', financed by a 'super-premium' uniformly applied to all insurance contracts (12% for home insurance and 6% for car insurance) (André, 2013).
The second result concerns compensation criteria in the event of expropriation. 29% of respondents felt that using market price provided the fairest basis for compensation. This criterion incorporates the libertarian principle (Nozick, 1974) of respect for property rights and the Utilitarian approach of maximizing total surplus, regardless of individual situations. Its importance, identified by our model, is in keeping with French practice. Currently in France, compensation payments are made with no consideration of the situation of individuals, whether in terms of prior information on the risk incurred or in terms of income. Although we cannot exclude the possibility that adhesion to national funding and market-price based compensation merely reflects current practice in France, our work brings in a new dimension by adding the responsibility principle to compensation criterion.
Finally, the study investigated the date of purchase of the assets and the number of years spent in the area as factors to be included in compensation criteria (Table 2). For owners who have longest lived in the area, this criterion gives value both to their attachment to their property and to their first-arrival rights, but it may also discriminate against owners who have purchased recently with knowledge of the risks. This solution introduces the responsibility principle. According to luck egalitarianism ideas, solidarity does not apply when inequality arises from factors that were deliberately selected by people who were responsible for their choices (Fleurbaey, 2008;Konow, 2001;Roemer, 1993). It was of particular interest to introduce the responsibility criterion into the proposed compensation mechanisms because since 2007, any risk is notified at the time of purchase in France. Results show that respondents who disagree with the idea that identical compensation for recent owners is unfair as they were informed of the risk incurred, tend to choose the status quo over managed realignment (Table 4). Hence, the third result of the study highlights that introducing responsibility into the compensation criteria may significantly increase the probability of respondents choosing managed realignment over the status quo. This confirms the tendency observed more generally in climate change where the responsibility principle plays a major part (Cooper & McKenna, 2008;Grasso, 2010;Miller, 2008;Paavola & Adger, 2006).

Conclusion
The study shows that the questions of funding realignment policies and of compensation criteria draw on different distributive judgements on the part of individuals. Whereas a principle of solidarity dominates in the case of funding, the results concerning compensation highlight the importance of two principles: the libertarian respect for property rights and the inclusion of individual responsibility of those who were informed of the risks.
Justice trade-offs usually relate to efficiency and equal opportunity principles according to various criteria. For example, in France and in Europe, generic surveys of these issues reveal in decreasing order the need, merit and equality criteria (Forsé & Parodi, 2006). But they also show that preferences vary according to the issue, depending in particular on whether economic growth or welfare is involved. It is difficult in our case to know the motivations of the respondents, given that managed realignment whilst maintaining beaches (and therefore improving the well-being of beach users) will also maintain the attractiveness of the coast for tourists and hence the local economy of these areas which are heavily dependent on tourism.
Finally, this work, carried out directly with the local population, shows that people's perceptions of justice and solidarity criteria are important to inform decision-making in shoreline management policies, where new reference points must be put into operation. Our research shows that the main two determinants of preferences and WTP relate to efficiency and utilitarianism in the sense that they introduce respectively a notion of solidarity towards those who have longest been owners and are most attached to their property and a notion of responsibility towards those who were informed of the risks. This result makes the study of distribute preferences a key element in enhancing the acceptability of environmental policies (Johansson-Stenman & Konow, 2010;Turner, 2007). In France, an experimental programme has been implemented in five pilot sites by the Ministry for Ecology and Sustainable Development to identify the regulatory and financial constraints to the implementation of these processes (MEDDE, 2012). However, few studies focus on residents' preferences for compensation or funding criteria. Hence, our approach paves the way for further analyses to explore, as Cai, Cameron, and Gerdes (2010), Klinsky, Downlatabadi, and McDaniels (2012) have done, various levels of compensation and responsibility.