Irish Case
Agent‑base modelling within a decision‑support system: Formulating policy for integrated rural tourism
Authors
Jean-Paul Bousset1 , Mary Cawley2 , Brenda Gallagher3 , Desmond A. Gillmor4
1UMR Metafort, Cemagref, Groupement de Clermont-Ferrand, 24 avenue des Landais, BP 50085, 63172 Aubière Cedex. e-mail: jean.bousset@voila.fr
2School of Geography and Archaeology and Environmental Change Institute, National University of Ireland, Galway, Galway City, Ireland. e-mail: mary.cawley@nuigalway.ie
3School of Education, National University of Ireland, Galway, Galway City, Ireland. e-mail: brenda.gallagher@nuigalway.ie
4School of Natural Sciences, Trinity College Dublin, Dublin 2, Ireland
Corresponding author
Mary Cawley, School of Geography and Archaeology and Environmental Change Institute, National University of Ireland, Galway, Galway City, Ireland. Tel: 353-91-492171; Fax: 353-91-495505; Email: mary.cawley@nuigalway.ie
Agent‑base modelling within a decision‑support system: Formulating policy for integrated rural tourism
Abstract
Agent‑based modelling (ABM) has gained increased attention in recent years as a method of incorporating multiple stakeholders in the management of natural resource systems. Few studies document applications in the context of rural tourism which is also characterised by complex actor systems and resource use. Bousset et al. (2007) provide a model in this regard which was followed in the research reported here. This paper presents the results of the application of ABM within a decision‑support system (DSS) as a method of modelling possible future scenarios for the promotion of more sustainable forms of rural tourism, defined as integrated rural tourism (IRT). The objective of the exercise was to identify the most appropriate ways of increasing the number of tourists who visit a region for the quality of its natural and cultural resources. There was broad agreement among five key stakeholder groups about the desirability of the event. The modelling exercise highlighted the diversity of policy approaches that may be advocated to attain this end, even when negotiation is entered into to seek to reduce conflicts. A scenario workshop, where representatives of the key stakeholders were presented with the results of the computer‑based analysis, revealed similar conflicts over resource use. The value of a modelling exercise, such as this, lies in the several possible future policy scenarios that it identifies and the conflicts that it reveals. Such outputs highlight areas which policy makers must address if they seek to attain a defined event.
Key words: Agent‑based modelling, decision‑support system, integrated rural tourism, Ireland
Agent‑base modelling within a decision‑support system: Formulating policy for integrated rural tourism
1. Introduction
Increased attention has been given in recent years by both academics and policy makers to finding more effective ways of managing natural resources in order to promote sustainability in their use (Jenkins, 2000; Parker et al., 2003; Saarinen, 2006; Capella et al., 2007). A particular challenge relates to including the multiple stakeholders who are often involved. Agent‑based modelling (ABM) provides a method of pursuing such inclusivity in decision‑making and has been applied in several recent studies relating to the planning of land and water usage (Grimble and Wellard, 1997; Mirchell et al., 1997). ABM is facilitated by being conducted within a decision support system (DSS) within which different possible scenarios are modelled (Bowler, 1999). Such an approach offers considerable potential for rural resource use, yet few geographical studies are available where the methodologies involved are described (Clifford, 2008). This paper seeks to contribute to the literature by describing an application of ABM within a DSS, using the example of promoting integrated rural tourism (IRT) based on natural and cultural resources in western Ireland. The concept of IRT is taken from Oliver and Jenkins (2003) who sought to promote a more holistic approach to pursuing sustainability in rural tourism which takes account of multiple stakeholders and resources.
The promotion of rural tourism is in essence a multi‑agent problematic. Tourism is allocated a role of growing importance in rural development policies nationally and internationally, in the wake of local and global economic restructuring (OECD, 1995; van der Ploeg, 2006). The rationale for such an approach is well founded because a growing demand exists for the traditional ways of life and outdoor recreational opportunities that characterise such environments among so‑called post‑modern tourists. The local comparative advantage resides in promoting niche high quality environmental and cultural experiences which draw on a wide range of natural, economic, cultural and social resources and which involve several stakeholder groups (Ryan, 2002; Northcote and Macbeth, 2006). Oliver and Jenkins (2003) identified seven distinctive features that should be prioritised. These relate to attaining multi‑dimensional sustainability and the empowerment of local people through appropriate use of local resources, local ownership, development of a scale appropriate to the context in which it takes place and complementarity between sectors. Effective networking, both at a horizontal level to increase the embeddedness of tourism and vertically, in a disembedded way, to reach external resources and tourist markets is a central element of such an approach (Saxena et al., 2007). Whilst the concept of IRT addresses issues of multidimensional sustainability, a well‑defined problem‑solving model is required which can specify the nature of the integrative linkages, the opportunities and benefits and the future development of IRT. Bousset et al. (2007) illustrate that these issues can be formally addressed by using an interactive and recursive modelling process and, specifically, by simulating models of decision-making in various hypothetical situations.
The questions to be addressed in the research discussed here were: to identify the nature of the integrative linkages in sustainable rural tourism; to formulate strategies for the use of resources most appropriate to optimising the utility functions of the regional actors involved, maximising the economic impacts and minimising the environmental impacts of tourism development; to identify institutional structures and strategies most appropriate for developing IRT initiatives; to identify the opportunities and benefits of IRT (including new product development and the exploitation of new markets) under various alternative future scenarios; to evaluate the consequences for resources and communities of IRT; and to model ways of resolving possible intraregional and intra‑community conflicts arising from IRT. These questions were addressed within the context of a hypothetical event which supposes that: “the quality of the natural and cultural environment becomes increasingly important to people generally, and the expectations of visitors to the study region about the quality of the natural and cultural environment strongly grow such that they are less inclined to tolerate its neglect. Consequently, actors’ concerns and strategies about the quality of the natural and culture resource base could change significantly in the near future”.
2. DSS and agent‑based modelling
A DSS offers some advantages over the more frequently used method of guiding the decision‑making of policy makers by asking experts what they think will happen (Turban, 1988). Experts rarely make explicit the criteria and the logic of combining the criteria that they use in making their assessments. Different experts also may have different suggestions. Largely in response to these shortcomings, policy analysts have begun to use statistical or econometric techniques to assess policy decisions (Bousset and Vollet, 2002). These approaches make explicit the logic that governs the selection of the independent variables that explain a phenomenon and the way in which they are combined. However, they are not wholly satisfactory because they do not make explicit the processes that govern the value of the dependent variables. Statistical forecasting derived from such methods tends to be accurate for stable trends but unreliable in forecasting significant changes in actors’ behaviour which are often of most interest to policy makers. Forecasting methods are needed that can both make explicit the logic that governs the choice of variables used to describe the behaviour and the logic that governs the behaviour (Bousset, 1994; Deffuant et al., 2003). A DSS meets these two objectives.
ABM is used increasingly at both local and regional scales to find more effective strategies for resource usage in multi‑agent systems (Parker et al., 2003; Grimble and Wellard, 1997; Mitchell et al., 1997; Capella et al., 2007; Clifford, 2008). ABM provides scope to incorporate the views and needs of multiple stakeholders, including those at a local level as well as centralised decision‑makers (Becu et al., 2008; Billgren and Holmén, 2008). Several applications of ABM are documented as a method of negotiating between many stakeholders in the context of land and water usage (Balman, 1977; Barreteau et al., 2003; Sperb et al., 2006; Gurung et al., 2006). Happe et al. (2006) illustrate the value of simulation models in examining various potential futures and therefore their superiority over isolated analysis of a policy regime, in the context of the potential impacts of the 2003 reform of the EU’s Common Agricultural Policy. O’Sullivan and Haklay (2000) suggest that many applications of ABM tend towards an individualist view of the world; in the present case the views of institutions, as expressed in policy preferences, are included also. The application of ABM in tourism contexts is not well documented in published literature, although more effective methods are needed to integrate rural tourism with the local physical and cultural resource bases on which it depends (Anwar et al., 2007; Bramwell and Lane, 2000).
Bousset et al. (2007) include a participatory and integrated policy process (PIP) and the use of multi‑agent simulation methodologies (MAS) as a method of identifying possible future policy options. The PIP approach to policy formulation brings together stakeholders and their activities from all relevant sources or sectors in ‘scenario workshops’ with a view to harmonising their conflicting objectives, strategies and capacities (Campbell and Townsley, 1997). These workshops are designed for the purpose of analysing pre‑constructed scenarios which describe the possible impacts of hypothetical events and policies. They may take the form of actual meetings between stakeholders, involving role playing games, which are usually conducted in a sequential way as a method of negotiating conflicts and attaining agreement on appropriate future policy options (Street, 1997). The negotiation process may also be simulated through specially‑designed computer systems. ‘Simulation’ involves running a model forward through various hypothetical contexts and times and watching what happens. Like deduction it begins with a set of explicit assumptions but unlike deduction it does not prove a theorem. Instead it generates data that are analysed deductively in order to understand a particular phenomenon better. Unlike typical deduction, the data come from a rigorously specified set of rules rather than from a direct measurement of the real world. While induction searches for patterns in data and deduction searches for consequences of assumptions, simulation just aids intuition (Gilbert and Doran, 1994). The simulation method includes five main stages (Doran, 1997): the definition of the questions to be addressed by the simulation of the model; the specification of the model; the application of the model in particular contexts; the verification of the model; the validation and interpretation of the outputs.
The use of PIP and MAS in a DSS is especially appropriate to the design of more effective policies for the promotion of integrated tourism in rural contexts. Rural tourism involves multiple stakeholders: business owners and managers, providers of resources for tourism, tourists, tour operators, host community members and representatives of institutions involved in policy and planning pertinent to the sector. These stakeholders use a wide range of natural, economic and socio‑cultural resources including land, buildings, finance capital, human capital resources and cultural attributes that are used to promote tourism. Effective policy design, with a view to attaining particular objectives for tourism in a regional or local context, should ideally reflect the interests and needs of the range of stakeholders. A DSS involving PIP and MAS provides a mechanism for facilitating negotiation between the stakeholders and simulating their actions under given scenarios in order to attain a hypothetical future event. The outputs of such an exercise should help to guide policy‑makers in planning for the future of IRT.
The context in which tourism evolves is influenced by both external and internal events over which stakeholders have varying levels of control. External forces, such as the events of September 11th 2001 on international air travel or the outbreak of Foot and Mouth Disease in the UK in spring 2001, in restricting access to land for recreational purposes, are outside the influence of local stakeholders. The ‘event’ analysed and reported here related to the potential provided by increased interest in natural and cultural tourism to promote this sector in western Ireland where nature and culture form part of its niche character. Internal forces that impinge on the future evolution of tourism in the region were hypothesised as relating to: (i) the resources and products available for tourism development (as hypothesised by the model); (ii) the resources, perspectives and strategies of the individual stakeholders involved in tourism development in the region; and (iii) the hypothetical coordination patterns governing the interactions among stakeholders and between stakeholders and local resources. The hypothetical coordination patterns evaluated included: (i) a policy supporting a greater supply of local tourism products; (ii) a policy supporting more promotion of local products and places; (iii) a policy promoting more cooperation between resource controllers, tourism businesses and tour operators to develop tourism; (iv) a policy supporting more provision of local resources; (v) a policy supporting more control of local resources; (vi) a policy supporting the creation of new regulations; (vii) no change in current tourism policies; (viii) a collaborative negotiation process among the different tourism stakeholders.
The scenario exercise involves evaluating the coordination pattern under different policies and the complex algorithms used to model these patterns are available from the first author. The evaluation based on current policies is known as the ‘baseline scenario’. Alternative scenarios were built by analysing the results of simulations of models of the collective behaviour of the stakeholders in face of the hypothetical event. Simulation consists of running modelled behaviours forward through various hypothetical contexts and times and seeing what happens. As a method of validating the results, the results of each scenario modelling exercise were presented to a group of representatives of key stakeholders involved in rural tourism in the region to elicit their responses relating to appropriate future strategy and policies for integrated tourism.
3. Methods
3.1 The modelling exercise
The application of each model includes four steps. The first step involves defining the stakeholder system. The second involves specifying the stakeholders’ resources, their perception of local resources and products, their goals and actions and, for all stakeholders (except tourists), their positions in face of the studied hypothetical event. These positions include the importance to them of the event, their expectations about the event, the actions that they would like to see done by others to achieve their expectations, their preferred actions, and the actions that they have already undertaken or plan to achieve. These data were collected through questionnaire surveys. The third step consists of analysing the data collected. This analysis indicates: (i) which stakeholder coalitions the studied event could generate in the future and the resources and strategy for each coalition (the expectations and preferred actions of their members); (ii) the consequences of these possible strategies and coalitions on the tourism value network and, thus, other possible conflicts or agreements that may occur among the stakeholders in the region in the future. These indications constitute the ‘base’ scenario. The fourth step consists of running the models which will indicate other possible strategies in face of the hypothetical event and thus other possible coalitions and conflicts amongst the stakeholders, as well as their effects on local resources, communities and tourism. These results are known as the ‘alternative’ scenarios.
The models were verified in two ways. This was done firstly by analysing the individual behaviours of two or three well known and also randomly selected stakeholders. Second, the simulation was run multiple times with varying values for some input parameters (e.g. the actors’ knowledge of their environment, their willingness to listen to and take account of the other actors) and by analysing the plots showing the changes in the outputs generated by these shifts. This analysis gives indications on the functional form of the relationships between the shifted input parameters and the outputs of the models and shows whether small changes in some inputs give rise to large changes in the outputs. An example relates to how much the actors’ strategies in face of the hypothetical event change whenever their willingness to listen to and take account of other actors is increased or decreased. The scenarios represent a possible future among many and it is not possible to validate the results with observed empirical data for the future. The results of the models are viewed as ‘scenarios’ (Godet, 2001). They are not predictions; the methodology does not suggest the probability of their occurrence or even their plausibility. They are ‘a’ future, not ‘the’ future. The interpretation of the scenarios took place during a participatory stage which involved their presentation to a panel of selected experts with thorough knowledge of the issues under discussion with the aim of developing visions and proposals for various needs and possibilities in the future. This approach is described as an interactive or deliberative method (Street, 1997). The objective is not to attain consensus but instead to generate discussion relating to a choice between different types of strategies where differences are important and where exchange of professional expertise may create new knowledge.
Two different models of behaviour were used: a ‘regulation’ model, which simulates stakeholder behaviours in the face of a specific policy (e.g., a policy supporting a greater supply of local tourism products); and a ‘negotiation’ model, which simulates a collaborative negotiation process among the different tourism stakeholders.
3.2 Regulation model
In the ‘regulation’ model, stakeholders have their own resources (including investment capability and workforce) and networks (ability to influence collective decisions through group membership and social relations). They are time independent and have preferred ways of achieving their expectations which have more or less important consequences for them. Each simulation of the decision‑making regulation model involves five stages: initialisation, provision, integration, promotion and consumption. The last four stages are sequentially and iteratively estimated ten times in order to simulate a long policy period. Initialisation creates a hypothetical space that represents the study area and reflects the spatial variation in resources and products existing among the stakeholders. The initialisation stage consists of two phases. The first consists of examining, with respect to the individual stakeholders, their resources, their expectations in terms of policy, their perception of the event (its importance to them) and of the potential of local resources for tourism development and their attitudes or behaviour towards issues relating to integrated tourism. The second stage involves allocating each stakeholder to a place (or ‘patch’) within the hypothetical space according to the resources they hold.
The provision, integration and promotion phases of the simulation process involve reallocating within the hypothetical space those resource controllers, businesses and gatekeepers who support the hypothetical event and believe that the tested policy is the best way to achieve it. As described above, six possible policy options or coordination patterns were tested, including no change in the current tourism policies. The provision phase relocates each resource controller who supports the event to places within the hypothetical space endowed with natural and cultural resources not yet available for tourism development. The integration phase relocates tourism businesses that conform to the above criteria and usually integrate natural and cultural resources into their products. They move to places in the hypothetical space endowed with resources accessible but not yet integrated into tourism products. The promotion phase relocates gatekeepers, who conform to the criteria, to places in the hypothetical space endowed with resources that are not yet promoted.
A screening process is used in the provision, integration and promotion stages to relocate stakeholders within the hypothetical space and is based on the stakeholders’ responses to specific questions in the questionnaires. The process includes three stages: censing, assessment, moving. The censing stage involves selecting the stakeholders who agree with tourism development and believe that the presented policy is the best way to achieve this goal. The assessment stage counts the number of tourists who visit the places where resources have been made more accessible, integrated and promoted by those stakeholders since the beginning of the simulation. If the number decreases, some supporters of the tested policy stop searching for new places to operate. If the number increases, some stakeholders who agree with tourism development but believed, hitherto, that the tested policy is not the best way to develop tourism become supporters of the policy and search for new places to operate.
The consumption stage simulates new tourists who come to the region, for its natural and cultural resources, exploring the region to satisfy their expectations in terms of natural and cultural resources, and social and economic resources. In doing so, their decision to stay in a particular place or move is determined by the number of other tourists present and the attitude of host community members towards the hypothetical event. At the end of the consumption stage, each resource controller, business and gatekeeper assesses the number of tourists who stay in the patches that they have made more accessible, integrated and promoted and they decide whether or not the particular strategy should be applied in the next run of the model.
3.3 Negotiation model
The ‘negotiation’ model aims to determine the strategies that the selected stakeholders should adopt if they wish to optimise their utility functions in face of a hypothetical evolution of their environment (i.e. an increase in the number of tourists who visit because of the quality of the natural and cultural resources of the region). In particular, it aims to identify the coalitions and conflicts that the event could generate among the local stakeholders. In doing so, it indicates possible future shifts within the strategies and partnerships that are implemented by the current stakeholders. This helps in achieving a better understanding of the potential roles of public and private stakeholders in the formulation of structures and policies to assist integrated tourism in the study region. There are three main assumptions underpinning the negotiation model: (i) the management of local resources is distributed among numerous public and private stakeholders with different but reciprocally dependent interests; (ii) in these circumstances, one way of mobilising and pooling stakeholders’ resources for the development of IRT consists of building and implementing specific networks known as ‘policy networks’; (iii) the construction of these policy networks can be constrained by a lack of shared vision for the future development of the region among local stakeholders, including host community members, by the absence of a culture of cooperation among private and even public stakeholders, and by the stakeholders’ knowledge which depends on communication between them, i.e. their networking.
Policy networks are defined as sets of autonomous stakeholders with different but reciprocally dependent interests (Knoke, 1990; Kenis and Schneider, 1991; Rhodes and Marsh, 1992; Konig and Brauninger, 1998), and a common problem‑solving orientation is reached through the compromise of interests and the emergence of trust in a stable network situation. Stakeholders are said to be reciprocally dependent when one stakeholder may prevent another stakeholder from achieving one of his/her goals (Castlefranchi et al., 1992). In the modelling exercise, policy networks were used as research tools to explore possible future informal structures able to induce cooperative behaviour between tourism businesses and between institutions and businesses, by simulating a utility maximisation model (Stokman and Zeggelink, 1996). The negotiation model is derived from a social exchange model which incorporates the stakeholders’ networks and their access relations (an access relation arises when the request of one stakeholder is accepted by one other stakeholder) (Stokman and Zeggelink, 1996). In the model, network formation is connected to the formation of the stakeholders’ positions, which is one of the main characteristics influencing the process. First, the model assumes that the stakeholders have their own preferred value for a given feature of their working environment (e.g. for an increase in the number of tourists who visit the region for its natural and cultural resources) to which they attach various importance, in their preferred way, to achieve this value. At the beginning of the simulation, these expectations, perceptions and preferences are derived from responses given to specific questions during the questionnaire surveys, as for the regulation model. Second, the negotiation model assumes that the stakeholders seek to influence each other’s positions through access relations in order to obtain policy outcomes as close as possible to their own positions. Therefore, access relations are instrumental in optimising utility functions. Maintaining access relations requires time and resources and the model assumes that stakeholders can request and accept only a limited number of access relations. Even for a small social system, stakeholders are unable to calculate the effects of access relations on the outcome of decisions and estimate their likelihood of success. The model therefore assumes that the stakeholders use simple heuristics and adapt their behaviour by learning retrospectively.
The heuristics represent two different views on politics. The first view is that stakeholders are primarily power driven and aim to have access relations with the stakeholder who has the greatest amount of resources in the field. Resources include investment capacity, workforce, ability to influence collective decisions (membership of a group), and social relations (direct contacts). Resources were estimated from the surveys. Stakeholders estimate their chances of success by comparing their own resources with those of the target stakeholders. Power also determines the order in which stakeholders accept requests. The second view is that stakeholders are primarily policy driven. In this view, stakeholders roughly estimate the effect that an access relation might have on the outcomes of decisions and accept requests selectively in order to bolster their own preferences as much as possible. Such a combined analysis of power, communication and policy stance reveals the importance of selective exposure to information. The negotiation model combines the two views and assumes that stakeholders are power and policy driven. In the negotiation model the highest goal of the stakeholders is assumed to be the attainment of policy outcomes that are as close as possible to their own preferences about both the state of the event and the actions to be undertaken to achieve that objective.
In the negotiation model each stakeholder has only limited resources. Consequently they have to choose which access relations they want to establish. An access relation from stakeholder i to stakeholder j is created if a request for access by i is accepted by j. The network of access relations is established by running the model many times. Each iteration comprises three steps: access evaluation and request, request evaluation, and position modification. The negotiation model provides three main outputs: (i) possible future changes among the current stakeholders’ coalitions (policy networks), i.e. new members, new objectives and changes in the preferred actions; (ii) the dynamics of these changes, i.e. the most influential stakeholders (who are able to challenge the greatest number of other stakeholders to change their positions) and the ‘brokers’ (stakeholders who can be challenged by other stakeholders to change their initial positions and who are able to challenge other stakeholders to adopt their new positions as well); (iii) the sensitivity of these two results to variations in the willingness of stakeholders to listen to others and in the relative importance of the stakeholders’ involvement in the current networks in the calculation of the stakeholders’ resources. Four alternative scenarios are built depending on willingness to listen to others (high willingness or low willingness) and the relative importance of involvement in networks (high involvement or low involvement).
4. Study area and stakeholders
The study area is a long‑established tourism destination in western Ireland. Tourism is heavily dependent on the region’s natural and cultural resources and the mountainous and coastal landscape, rivers and lakes feature strongly in promotional materials. Government policy has promoted tourism development more actively over the last two decades as a method of supplementing declining incomes from farming and fishing (DAST 2003). Significant investment has taken place in the region in tourism‑related infrastructure (roads, hotels, interpretative centres, golf courses, marinas) (Deegan and Dinneen, 2000). Much of this investment has come from the private sector with state support. The number of tourists visiting the region has increased significantly over the past decade, although the share of national tourism income remains at approximately 15% (Deegan and Moloney, 2005). Touring by car and coach is well‑established in the coastal zone and synergies exist between urban and rural areas. Nevertheless, considerable concentration of expenditure and investment takes place in selected urban and rural locations and more remote or less scenic rural areas remain less well integrated into the nexus of tourism. Among the challenges for policy makers are, inter alia, to attract tourists who are interested in the characteristic natural and cultural features of the region, to devise ways of maintaining the quality of those resources, to prevent over‑crowding in particular locations and pressure on resources, and to spread the tourists more widely to less frequented areas. The MAS model provides a method for negotiating policies to achieve these aims.
Six groups of stakeholders were interviewed: 51 tourism business owners and managers, 23 providers of cultural, economic, natural, social and physical resources for tourism, 11 gatekeepers (tour operators), 50 host community members, 20 institutions with roles concerning policy and planning for tourism, and 110 tourists. Local, regional, national and international levels of operation were represented as appropriate. The interviews collected information relating to: (i) the individual stakeholders’ resources (employee numbers, turnover, and number of tourists dealt with, as pertinent to the stakeholder group) and the presence or absence of current networking with others (membership of [other] organisations or groups, including membership of Boards, and joint initiatives with such bodies); (ii) their views and perceptions with reference to the importance of local resources for tourism and the importance of the scenario event in relation to their own operation; (iii) their expectations relating to the future changes in conditions surrounding tourism when present conditions are taken into account and the ways in which resources should be used to support these expected changes; and (iv) their attitudes or behaviour relating to a series of issues relating to integrated tourism.
5. The DSS in operation
5.1 The base scenario
The resources at the disposal of the various stakeholders have implications for their aspirations and attitudes relating to the type of tourism that is most appropriate for the study region. They also influence the stakeholders’ capacity to negotiate with each other to maintain current policy or promote a change in policy. On average the gatekeepers and institutions in the study region were more resourced than were the resource controllers and the businesses (with an average resource index of 0.4 and 0.2 versus 0.1 and 0.1, respectively). The businesses and the institutions were, however, more involved in networking as compared to the gatekeepers and the resource controllers (with average networking indexes of 1.96 and 1.95 as compared with 1.00 and 0.83, respectively). These features suggest that potential influence is somewhat dispersed between the different stakeholder groups with no individual group dominating.
The base scenario describes the individual reactions and policy expectations of the stakeholders in face of the hypothetical event, as expressed in the questionnaires. It also hypothesises the implications of these policies for the number of new tourists who will come to the region and stay there, based on conflicts with host community members. The majority of the stakeholders (86%) considered the tourists who come to the study region for the quality of its natural and cultural resources to be very important (they are defined as ‘supporters’ of the event) (Table 1). The event was assigned greatest importance by the institutions, all of whom indicated that it was very important, and assigned least importance by the resource controllers, 13% of whom said that it was of no importance. The large majority of stakeholders (91%) would like to see an increase in the number of tourists who come to the region for the quality of its natural and cultural resources (Table 2). Institutions were the main supporters. Gatekeepers, resource controllers and host community members were those most likely to wish to maintain the current number of tourists (‘moderators’), whilst 4% of the resource controllers wished to see a decrease in the number of tourists who come to the region for the quality of its natural and cultural resources (‘opponents’).
5.2 Alternative scenarios from simulations of the regulation model
Notwithstanding the broad support for the event, there was no consensus among the supporters about the policies to be used to attain their preferences (Table 3). Policies that focused on the provision of resources, the control of resources or the promotion of local tourism products and places were most popular. About 38% of the gatekeepers, 36% of the host community members, 35% of the resource controllers, 26% of the businesses and 25% of the institutions, who have preferences about the ways to achieve an increase in the number of specified tourists, would like to see this policy focus on resource provision. Similar respective proportions of the gatekeepers (38%) and of the businesses (26%) would like to see a policy that focuses on the control of resources, and 25% of the institutions favour a policy that focuses on the promotion of products and places.
Consequently, the proportion of the local resources that the actors might make more developed, integrated and promoted (and finally visited by the new tourists) in the future, is sensitive to the targets and actions of future tourism policy. This is due to the variations existing among the actors’ preferences about the outcomes and actions of that policy, and to variations among the resources they can mobilise (Table 3).
Table 4 presents the results of the ‘regulation’ model with different policy options.
The promotion oriented policy and the provision oriented policy make more ‘visited’ (i.e. increase the number of tourists visiting) a greater proportion of the local resources as compared to the other single oriented policies. However, they induce a great deal of conflict between the host communities and the tourists. The supply oriented policy induces the lowest number of conflicts between the host communities and the tourists, but, due to the low proportion of the resources made more integrated and promoted by the adopters, a large proportion of the new tourists who come to the area for its natural and cultural resources are left unsatisfied. The control oriented policy, as with the market‑like evolution (no policy), makes more ‘visited’ a larger part of the resources but a substantial proportion of the tourists are left unsatisfied. In addition it induces a large number of conflicts. Hence the control oriented policy and the market-like evolution, are the least favourable options. The simulated regulation process, which puts all the above policies together, makes more ‘developed’, ‘integrated’, ‘promoted’, and finally ‘visited’ the greatest proportion of the resources. However, it may generate a great deal of conflicts among the stakeholders.
The preference for resource provision, resource control and product/place promotion oriented policies among supporters of the event are bases for alliances among these stakeholders based on their capacity to mobilise resources and their networks (Table 3). Gatekeepers, institutions and businesses would be strongly represented in such alliances, in particular among those with a preference for product/place promotion, resource provision and resource control policies. However such preferences are also bases for conflicts, especially between supporters with a preference for either a product/place promotion oriented policy, a resource control oriented policy or a cooperation (collective actions) oriented policy, because of similarities in their ability to mobilise resources and/or similarities in the value of their networks (Table 3). Supporters of a resource provision oriented policy are best placed to operationalise their preferences and are most likely to influence other stakeholders because of their high levels of resources and networks. As such these stakeholders challenge other supporters, moderators and opponents to join them to influence tourism policy. In addition, because of the value of their combined resources and networks, supporters of the resource control oriented policy may challenge supporters of the cooperation (collective actions) and the regulation oriented policies to join them to disrupt the actions of the supporters of the resource provision and product/service supply oriented policies. However, if tourism evolves in this way conflicts will increase between new tourists and host community members and the model shows that approximately 60% of new tourists will not find places to meet their preferences and will leave the region.
5.3 Alternative scenarios from simulations of the negotiation model
Following the negotiation process, some changes might be expected in the support for the outcomes of future tourism policy (the event) and also in the policy preferences to attain the outcomes, reflecting variations in the willingness of stakeholders to listen to others and in the relative importance of the stakeholder resources.
A number of the few stakeholders who would wish the number of tourists to remain as they are (moderators) or those who would not wish numbers to increase (1 opponent) would join those who support the event (Table 5). One resource controller would leave the opponents’ coalition and join the supporters. One gatekeeper and two host community members would leave the moderators’ coalition, i.e. those who wish to maintain the number of tourists, and join the supporters. One host community member would leave the supporters’ coalition to join the moderators’ coalition. Following negotiation, the supporters’ coalition would marginally increase in size. By increasing the number of supporters (from 141 to 144), and hence their resources and networks, and decreasing the number of moderators (from 13 to 11), and hence their resources and networks, the negotiation process would reduce the potential for conflicts between the supporters of the event and the moderators (Table 5). Such reductions would be greatest in relation to conflicts between gatekeepers and host community members, gatekeepers and businesses, resource controllers and host community members, and resource controllers and gatekeepers (Table 6).
Some supporters would change their position relating to the actions to be undertaken to achieve the outcomes of future policy (their policy preferences) during the negotiation process also. Those supporting more control would make the greatest gains at the expense of a policy favouring more application of existing regulations and more promotion of tourism products and places. Following the negotiation process, supporters would therefore be marginally more likely to have a preference for resource control oriented policies (Table 3 and Table 7). The resulting change in the stakeholder resources and networking has implications for conflicts between those supporting different policy preferences. An increase in networking increases the success of the supporters of the supply and control oriented policies in challenging the supporters of, particularly, the promotion oriented policies but also of the application, provision, and cooperation oriented policies, plus those who currently have no opinion, to join them (Table 3 and Table 7).
The shifts in the stakeholders’ expectations about the objectives of future policy are marginally sensitive to the variation in the importance of their networking in the decision-making process and somewhat sensitive to the variation in their willingness to listen to the expectations and preferences of others, but only in relation to gatekeepers and host community members (Table 6). An increase in networking marginally increases the success of supporters in challenging two host community members and a gatekeeper who act as moderators to join them. An increase in willingness to listen marginally reduces the success of supporters in challenging a host community member to join them (he/she changes to moderator status). Overall the negotiation process would have little impact in changing the potential for conflicts about the actions to be undertaken to achieve the outcomes of the future policy (Table 6). A marginal increase in the potential for conflicts would be the expected outcome among resource controllers and institutions (intra‑conflicts), while a decrease would be expected among the other stakeholder groups. The negotiation process would also only impact marginally upon the potential for conflicts between the stakeholder groups (inter‑conflicts) and would overwhelmingly result in a decrease in the potential for such conflicts. These deceases would be greatest between gatekeepers and businesses, gatekeepers and institutions, institutions and resource controllers and institutions and businesses.
5.4 Experts’ views
The outcomes of the base regulation scenario and the alternative scenario, in terms of policy preferences and conflicts, were presented to a group of fourteen well‑informed experts who represented five of the stakeholder groups (excluding the tourists) in a scenario workshop. Their responses to the preferences of the stakeholders in the regulation scenario and their own recommendations for attaining the event and reducing conflicts reflected the outcome of the modelled scenario closely. Most were supporters of an increase in the number of tourists who visit the region for the quality of its cultural and natural resources. There was a basic dichotomy in the actions favoured to attain their preferences between controllers of landscape and game fisheries resources and a landowner host community member and the other stakeholders. The former proposed more control of resources as a method of controlling tourist numbers, whereas other host community members, business owners, gatekeepers and institutional representatives recommended more promotion, more investment in transport infrastructure to improve access to the region, and more provision of indoor recreational facilities in remote areas to increase tourist numbers. A number of institutional representatives suggested making underutilised resources in remote locations more available. A gatekeeper suggested more networking between resource controllers and gatekeepers as a method of directing tourists to such areas.
In response to the outcomes of the negotiation process, a resource controller with responsibility for angling waters and a host community member said that some resource controllers could become more entrenched in their opposition to the event. They referred in particular to owners of private property, who were reluctant to provide access for hiking, and to managers of sensitive landscapes and fisheries waters. Views were sought on the actions to be taken to challenge such actors to make resources more available to the targeted tourists and the structures and policies required to promote the actions. The proposed actions included management agreements with landowners and acceptance of the importance of environmental conservation in promoting sustainable resources for future tourism. In terms of structures and policies to promote the actions it was suggested that management agreements to provide access to private property should be developed at a national level and implemented locally (which has taken place since the study was completed). It was suggested also by an institutional representative, a resource controller and a gatekeeper that businesses and host community members should become involved in management agreements with resource controllers locally in order to develop footpaths in a sustainable way. Educational programmes in schools, in the public arena and specific programmes for farm families were suggested by several stakeholders as methods of increasing awareness of the longterm benefits of environmental conservation for tourism. Until such awareness becomes apparent, some resource controllers would have difficulty in making further resources available to the tourists who might visit for the natural and cultural resources of the region who may therefore be disappointed and fail to return. Whilst disagreement remained between the stakeholders, the scenario meeting did help to identify policies to help attain the event and reduce conflicts.
6. Discussion and conclusions
The promotion of rural tourism involves state agencies, voluntary organisations and private entrepreneurs managing a wide range of resources in collaborative ways. A need for greater integration between stakeholders and resource uses is recognised (Oliver and Jenkins 2003). Policy-makers are interested in the future outcomes of strategies under different operational conditions rather than in the present or the past. Bousset et al. (2007) illustrate how possible future scenarios can be modelled in the context of a DSS which includes a participatory approach (PIP) and a multi-agent simulation exercise (MAS). Multi‑agent modelling of this kind is being applied increasingly in the context of land and water uses but is less well developed in the case of tourism. Rural tourism characteristically involves multiple resources and stakeholders, and finding a method of representing these various resources and interests is particularly important in the context of promoting holistic sustainability in tourism. This paper discussed the experience of applying a DSS to model IRT promotion in the context of western Ireland. The outcomes of the modelling exercise have direct implications for the promotion of sustainable tourism in the study area but also raise broader issues relating to the value of a DSS in negotiating between the interests of multiple stakeholders in order to inform public policy more effectively.
The underlying logic of the modelling exercise related to identifying appropriate policies to promote the sustainable use of endogenous natural and cultural resources in complementary ways, on an appropriate scale, so as to attract the desired types of tourists and contribute to the empowerment of local people. This required identifying preferred approaches to resource use and exploring the potential for reducing conflicts through a process of networking and listening to others. The exercise revealed that there was broad agreement with the event (an increase in the number of tourist visiting the region for the quality of its endogenous natural and cultural resources) but that considerable disagreement was present among the non‑tourist stakeholders relating to the optimum method of attaining the event. This disagreement reflected the inherent complexity of the stakeholder system and the varying objectives, resources and networking that were present, even within particular stakeholder groups, which are recognised features of rural tourism. The simulated regulation scenario modelled how stakeholders seek to maximise their utility functions, taking into account their capacity to influence others based on their resources and networking. This exercise indicated that support would increase marginally for the event and that some stakeholders would be willing to modify their policy preferences in order to attain the event. Disagreement remained, however, on the optimum policy for achieving the desired outcome. Simulation of a negotiation process, based on networking and a willingness to listen to others, illustrated that minor changes would result in support for various policies. A policy that favoured increased provision of resources continued to dominate because of the relatively wider support that it attracted across all stakeholder groups, but the number of stakeholders who advocated increased control of resources increased as did their relative influence.
The scenario workshop at which representatives of key stakeholders discussed the outcomes of the modelling exercise at length reflected similar disagreement on policy preferences to that in the modelled scenario. The responses to the scenarios presented during the workshop therefore served to validate the outcome of the simulation exercise. They also suggested methods by which some of the conflicts might be reduced. The business, host community and institutional representatives at the workshop sought methods of convincing the resource controllers to make scarce and under-used resources available (e.g. through access agreements to land and management agreements to prevent erosion on footpaths). Raising public awareness of the links between environmental conservation and tourism was advocated also. However, members who favoured increased control possess considerable power because of their resources and networking (e.g., regional fisheries board and regional wildlife and park service managers) and if additional tourists who seek these resources are attracted to the region, through investment in promotion and access infrastructure, they may not find experiences to meet their requirements and may not return. Discussion at the workshop suggested that the introduction of public educational and awareness raising programmes, relating to the need for environmental conservation, may influence some resource controllers to make underutilised but potentially sensitive resources available.
PIP and MAS expand on more traditional approaches to forecasting appropriate policy for resource use, based on expert opinion or econometric techniques, by taking into account the interests of multiple stakeholders and the logic of their decision-making in simulating possible future scenarios. Such stakeholder diversity is often characteristic of rural resource ownership. AMB within a decision‑support system provides a method of pursuing a more inclusive approach to designing effective policies in such contexts. It does not provide ‘a’ solution but instead reveals possible solutions, which can then form a basis for further discussion and ultimately policy‑making. Such information is of growing importance for the management of rural resources, the effectiveness of which depends on negotiating solutions between diverse stakeholder interests. The results presented here suggest that, ideally, both computer modelling based on survey data and participative scenario workshops should be combined in seeking solutions. Where extensive data collection may not be possible, multiple workshops with stakeholders may be used as a method of including stakeholder interests in policy design (Street, 1997; Gurung et al., 2006). The spatial remit of such exercises is, however, generally more limited in scope than that afforded by modelling based on comprehensive questionnaire data collection and its validity may therefore be more circumscribed.
A light and easy-to-use version of negotiation model can be found in https://prospector0.webnode.fr/products/negotiation-model/
Acknowledgement
The reported research was conducted as part of a six country EC-funded project, contract number 2000-01211.
References
Anwar, S.M., Jeanneret, C.A., Parrott, L. and Marceau, D.J., 2007. Conceptualization and implementation of a multi‑agent model to simulate whale‑watching tours in the St. Lawrence Estuary in Quebec, Canada. Environmental Modelling and Software 22, 1775-1787.
Balmann, A., 1997. Farm‑based modelling of regional structural change: A cellular automata approach. European Review of Agricultural Economics 24, 85-108.
Barreteau, O., Garin, P., Dumontier, A., Abrami, G. and Cernesson, F., 2003. Agent‑based facilitation of water allocation: Case study in the Drome River Valley. Group Decision and Negotiation 12, 441-461.
Becu, N., Neef, A., Schreinemachers, P. and Sangkapitux, C., 2008. Participatory computer simulation to support collective decision‑making: Potential limits of stakeholder involvement. Land Use Policy 25, 498-509.
Billgren, C. and Holmán, H., 2008. Approaching reality: Comparing stakeholder analysis and cultural theory in the context of natural resource management. Land Use Policy 25, 550-562
Bousset, J-P. 1994. Decision-making process and strategic planning process by scenarios: An operative model to study the possible futures of cattle and sheep farms of the Auvergne and Limousin. Farmer’s Decision Making, 38th EAAE Seminar, Copenhagen, October 1994.
Bousset, J‑P. and Vollet, D., 2002. Use of meta‑analysis for the comparison and transfer of economic base multipliers. Regional Studies 36, 481-494.
Bousset, J‑P, Skuras, D., Tesitel, J., Marsat, J‑B., Pertou, A., Fallo‑Pantziou, E., Kusova, D. and Bartos, M., 2007. A decision support system for integrated tourism development: Rethinking tourism policies and management strategies. Tourism Geographies 9, 387-404.
Bowler, I., 1999. Modelling farm diversification in regions using expert and decision support systems. Journal of Rural Studies 15, 297-305.
Bramwell, B. and Lane, B., 2000. Tourism Collaboration and Partnerships: Politics, Practice and Sustainability. Channel View Publications, Clevedon.
Campbell, J. and Townsley, P., 1997. Participatory and integrated policy processes in small‑scale fisheries. PLA Notes 30, pp. 66-69. IIED, London.
Capella, J‑C, Kam, S.P., Quang, D.D., Verburg, P.H. and Hoanh, C.T., 2007. Combining top‑down and bottom‑up modelling approaches to land use/cover change to support public policies: Application to sustainable management of natural resources in northern Vietnam. Land Use Policy 24, 531-545.
Castlefranchi, C., Micelli, M. and Amedeo, C., 1992. Dependency relations among autonomous agents. In: Werner, E. and Demazeau, Y., Editors, Decentralized Artificial Intelligence 3. Elsevier, Amsterdam, pp. 215-227.
Clifford, N.J., 2008. Models in geography revisited. Geoforum 39, 675-686.
DAST (Department of Arts, Sport and Tourism), 2003. New Horizons for Irish Tourism. DAST, Dublin.
Deegan, J. and Dinneen, D., 2000. Market services and tourism. In: O’Hagan. J.,
Editor, The Economy of Ireland: Policy and Performance of a European
Region. 8th ed., Gill and Macmillan, Dublin, pp. 286-307.
Deegan, J. and Moloney, R., 2005. Understanding the Economic Contribution of Tourism to Economic Development: The Case of Ireland West. Ireland West Tourism, Galway.
Deffuant, G., Huet, S., Bousset, J-P, Henriot, J., Amon, G. and Weisbuch, G., 2003. Agent‑based simulation of organic farm conversion in Allier department. In: Janssen, M.A., Editor, Complexity and Ecosystem Management: The Theory and Practice of Multi-Agent Systems. Edward Elgar, Cheltenham, pp. 158-187.
Doran, J., 1997. From computer simulation to artificial societies. Transactions of the Society for Computer Simulation International 14, 69-77
Gilbert, N. and Doran, J. (Editors), 1994. Simulating Societies: The Computer Simulation of Social Phenomena. UCL Press, London.
Godet, M., 2001. Creating Futures: Scenario Planning as a Strategic Management Tool. Economica, London.
Grimble, R. and Wellard, K., 1997. Stakeholder methodologies in natural resource management: A review of principles, contexts, experiences and opportunities. Agricultural Systems 55, 173-193.
Gurung, T. R., Bousquet, F. and Frébuil, G., 2006. Companion modelling, conflict resolution, and institution building: Sharing irrigation water in the Lingmuteychu Watershed, Butan. Ecology and Society 11, pp. 36. URL: http//www.ecologyandsociety.org/vol11/iss2/art36/. Downloaded 16 June, 2008.
Happe, K., Kellermann, K. and Balmann, A., 2006. Agent‑based analysis of agricultural policies: An illustration of the agricultural policy simulator AgriPoliS, its adaptation and behavior. Ecology and Society 11, pp. 49. URL: http//www.ecologyandsociety.org/vol11/iss1/art49/. Downloaded 16 June, 2008.
Jenkins, T., 2000. Putting postmodernity into practice: Eendogenous development and the role of traditional cultures in the rural development of marginal regions. Ecological Economics 34, 301-314.
Knoke, D., 1990. Political Networks: The Structural perspective. Cambridge University Press, Cambridge.
Kenis, P. and Schneider, V., 1991. Policy networks and policy analysis: Scrutinizing a new analytical toolbox. In: Marin, B., Mayntz, R., Editors, Policy Networks: Empirical Evidence and Theoretical Considerations. Campus Verlag, Frankfurt, pp. 25-29.
Konig, T. and Brauninger, T., 1998. The formulation of policy networks. Journal of Theoretical Politics 10, 445-471.
Mirchell, R.K., Agle, B.R. and Wood, D.L., 1997. Toward a theory of stakeholder identification and salience: Defining the principle of who and what really counts. Academy of Management Review 22, 853-886.
Northcote, J. and Macbeth, J., 2006. Conceptualizing yield: Sustainable tourism management. Annals of Tourism Research 33, 199-220.
Oliver, T. and Jenkins, T. 2003. Sustaining rural landscapes: The role of integrated tourism. Landscape Research 28, 293-307.
OECD (Organization for Economic Co-operation and Development), 1995. Niche Markets as a Rural Development Strategy. OECD, Paris.
O’Sullivan, D. and Haklay, M., 2000. Agent‑based models and individualism: Is the world agent‑based? Environment and Planning A 32, 1409-1425.
Parker, D.C., Manson, S.M., Janssen, M.A., Hoffmann, M.J. and Deadman, P., 2003. Multi‑agent systems for the simulation of land‑use and land‑cover change: A review. Annals of the Association of American Geographers 93, 314-33.
Rhodes, R.A.W. and Marsh, D., 1992. New directions in the study of policy networks. European Journal of Political Research 21, 181-205.
Ryan, C., 2002. Equity, management, power sharing and sustainability: Issues of the ‘new tourism’. Tourism Management 23, 17-26.
Saarinen, J., 2006. Traditions of sustainability in tourism studies. Annals of Tourism Research 33, 1121-1140.
Saxena, G., Clark, G., Oliver, T. and Ilbery, B. 2007. Conceptualising integrated rural tourism. Tourism Geographies 9, 347-370.
Sperb, R.M., Cabral, R.B. and Bughi, C.H., 2006. Fuzzy agents: A hybrid tool for exploring coastal zone spatial processes. Journal of Coastal Research 3 (Special Issue 39), pp. 1510-1514.
Stokman, F.N. and Zeggelink, F.P.H., 1996. Is politics power or policy oriented? A comparative analysis of dynamic access models in policy networks. Journal of Mathematical Sociology 21, 77-111.
Street, P., 1997. Scenario workshops: A participatory approach to sustainable urban living? Futures 29, 139-158.
Turban, E. 1988. Decision Support and Expert Systems. Macmillan, New York.
van der Ploeg, J., 2006. Agricultural production in crisis. In: Cloke, P., Marsden, T.
and Mooney, P., Editors, Handbook of Rural Studies. Sage, London, pp. 258-77.
*1 Gatekeeper who had ‘no preference’ in the base scenario supported more provision of resources following negotiation.
B=business owner/manager; HC=host community member; I=institutional representative; RC=resource controller