مدل ها برای سیاستگذاری در توسعه پایدار: وضعیت هنر و دیدگاه برای تحقیق

More and more frequently policy-makers are urged to assess the impact of their strategies and policies in terms of sustainable development. This necessitates the use of applied scientific models as tools for identifying and evaluating the likely environmental, economic and social impacts of alternative policies. The objective of this paper is to provide a framework to help decision-makers choose the most appropriate—or the most appropriate mix—of models, by assessing their relative strengths and weaknesses. The paper also allows potential improvements in modeling techniques to be identified. Six modeling paradigms are assessed, both on a general basis and with respect to two specific policy contexts (energy policy, and land use and transport planning).

More and more, policy-makers are urged to assess the impact of their strategies and policies in terms of sustainable development2. So much so that an—allegedly—new discipline named ‘Sustainability Impact Assessment’ (SIA) has been created to address these issues (Lee and Kirkptrick, 2000 and Lee and Kirkpatrick, 2001). A crucial stage in SIA is anticipating the likely economic, environmental and social impacts of the planned policy. For long-term and complex policy matters, this is only feasible with mathematical or computer-based models. However, there are several different approaches to economic–environmental or integrated modeling and it is not easy for the policy-maker to decide which is the most appropriate for any context. Our objective here is to help users to choose the most suitable modeling tool for a particular sustainable development problem and to better understand what kind of information can be expected from the models. This issue has been remarkably neglected in the literature. The only paper addressing a similar question is that by van den Bergh and Nijkamp (1991) in this journal, but there is no standard procedure for evaluating the strengths and weaknesses of different modeling approaches for sustainable development policy-making. Our contribution is to elaborate a formal methodological framework to tackle this issue and to apply it to existing modeling paradigms and two policy fields. This problem is a decision-making one. It has to do with the identification of the possible alternatives (the various modeling approaches and tools), the selection of criteria by which to assess them, the assessment itself with respect to the criteria, the weighting of the criteria and, finally, the aggregation of the partial assessment (on each criterion) in an overall assessment. This is, more or less, the way we will proceed in this paper. Several modeling approaches will be assessed in two stages: first with respect to general criteria closely related to sustainable development and then in relation to policy matters (energy and land use and transport policies) considered from a sustainable development perspective. Six modeling paradigms will be assessed, first on a purely a priori and general basis, and then against two specific policy contexts (energy policy, and land use and transport planning). The paper is organized as follows. We begin by stressing what is specific in sustainable development in order to decide on the most relevant assessment criteria. The modeling paradigms are then compared against these criteria on a purely a priori basis and ranked with respect to their potential performance in dealing with sustainable development problems. The robustness of this ranking is then checked in two concrete policy contexts: energy policy, on the one hand, and land use and transport policy, on the other hand. These policy domains are considered as collections of still more concrete issues (such as resource exhaustion and energy dependency for the energy case), each embodying the essence of sustainability at different levels. The relative fitness of the various modeling approaches to these policy domains is considered as a function of: (i) the degree to which the policy domains embody sustainable development characteristics; and (ii) the degree to which the modeling paradigms are able to deal with these characteristics. Finally, we look at existing modeling practices in the two policy fields in order to see if they confirm our conclusions about the usefulness of the different modeling paradigms.

As far as we know, there is at present no standard procedure for evaluating the strengths and weaknesses of different modeling approaches for sustainable development policy-making. The modus operandi adopted here was loosely inspired by multi-attribute utility theory (MAUT). In short, MAUT consists in selecting a set of relevant criteria (attributes), giving each criterion an importance weighting and then rating each alternative with respect to the criteria. If there are many decision-makers, the individual ratings are then aggregated. Finally, the ratings are once again aggregated (by addition, multiplication or whatever) across the various criteria weighted by their relative importance. Here, we have drawn attention to five methodological attributes of equal importance which are central to sustainable development decision-making: interdisciplinary potential, long-term and intergenerational concern, uncertainty management, local−global interaction, and stakeholders' participation. The following modeling paradigms have been scrutinized: macro-econometric models, computable general equilibrium models, optimization models, system dynamics models, multi-agent simulation models and Bayesian network models. All have been rated on an ordinal scale during participatory workshops attended by model users and builders. What is innovative is the way we then proceeded to assess the potential of the six alternatives for policy decision-making in energy and land-use issues. What emerges is that not all modeling approaches are equally helpful for sustainable development policy-making. Most importantly, we provide a rationale for our ranking. Naturally, every model has its own utility, and the one-size-fits-all model will never exist. What is important is to understand both how the models could be improved for sustainable development purposes and used in support of decision-making. In this respect, the models considered in this paper can be split with respect to their degree of involvement in the decision-making process to date. The evidence that the most intensively used are not the most suitable for sustainable development purposes suggests twin channels for research: firstly, how to use the best-performing models in the decision-making process; and, secondly, how to improve the goodness-of-fit of the modeling tools that are currently used. Our methodology may help in identifying the key features for further research. Unambiguously, the most promising modeling approach seems to be the multi-agent simulation model. It has many potential strengths to commend it. First of all, such models bypasses most mathematical jargon and simulate scientific hypotheses or even commonsense knowledge directly, without prior mathematical translation. Second, they allow for an intuitive representation of the environment and of the embedding of agents in a spatial and natural setting. Finally, they really display a ‘bottom-up’ structure, thus allowing an adequate representation of micro/macro relationships. Admittedly, multi-agent modeling represents a new paradigm and many theoretical and methodological problems remain to be resolved before it can be used on a regular basis for practical sustainable development policy-making. However, several powerful and user-friendly computer software for building agent-based systems are already available, some of them for free (e.g. Ascape28, Netlogo29 and RePast30). It is our opinion that public scientific and R and D policy-makers and advisers should foster their development and use in universities, schools and research institutions. Bayesian networks and system dynamics should also be more widely diffused. What makes them attractive from a sustainable development point of view is the fact that they combine an intuitive graphical user interface with an interdisciplinary or trans-disciplinary scientific language (Bayesian or subjective probability theory for Bayesian networks and general system theory for system dynamics). The graphical interface helps to translate stakeholders' knowledge and beliefs into workable equations or statements, while the universal language fosters dialogue between different scientific disciplines. To conclude, what is common to Bayesian networks, multi-agent simulations and system dynamics models that makes them relatively well-suited to sustainable development is their potential for cognitive integration, i.e. the integration of various kinds of knowledge, various scientific disciplines, different time-spans and different institutional and ontological levels. Such integration typically underpins modeling tools which are suitable for decision-making in sustainable development.