Multi-criteria analysis in environmental decision-making
Projects, policies and their impacts are embedded in a system of broader (national) objectives. If the impacts of projects and policies on these broader objectives can be valued economically, all such effects may be incorporated into the conventional decision-making framework of cost-benefit analysis (CBA). However, some social and biophysical impacts cannot be easily quantified in monetary terms, and multi-criteria analysis offers a complementary approach, which facilitates decision making.
Multi-criteria analysis (MCA) or multi objective decision-making differs from CBA in three major areas[1]. While CBA focuses on efficiency (although incorporation of income distribution objectives may be attempted), MCA does not impose limits on the forms of criteria, allowing for consideration of social and other forms of equity. Secondly, while CBA requires that effects be measured in quantitative terms, to allow for the application of prices, MCA can be broken down into three groups: one that requires quantitative data, a second that uses only qualitative data, and a third that handles both simultaneously. Finally, MCA does not require the use of prices, although they might be used to arrive at a score. CBA uses prices which may sometimes be adjusted according to equity weighting. MCA uses weighting involving relative priorities of different groups as opposed to pricing. If efficiency is the only criterion and prices are available to value efficiency attributes, CBA is preferable. However, in many cases, a paucity of data, and the need to incorporate social and biophysical impacts makes the use of MCA a more practicable and realistic option.
MCA calls for desirable objectives to be specified. These often exhibit a hierarchical structure. The highest level represents the broad overall objectives (for example, improving the quality of life), often vaguely stated and, hence, not very operational. They may be broken down into more operational lower level objectives (e.g., increased income) so that the extent to which the latter are met may be practically assessed. Sometimes only proxies are available (e.g., if the objective is to enhance recreation opportunities, the number of recreation days can be used). Although value judgments may be required in choosing the proper attribute, measurement does not have to be in monetary terms (like the single criterion CBA). More explicit recognition is given to the fact that a variety of concerns may be associated with planning decisions.
An intuitive understanding of the fundamentals of multi-criteria analysis can be provided by a two dimensional graphical exposition such as in Figure 1. Assume that a scheme has two non commensurable and conflicting objectives, Z1 and Z2. For example, Z1 could be the additional project cost required to protect biodiversity, and Z2, some index indicating the loss of biodiversity. Assume further that alternative projects or solutions to the problem (A, B and C) have been identified. Clearly, point B is superior (or dominates) to A in terms of both Z1 and Z2 because B exhibits lower costs as well as biodiversity loss relative to A. Thus, alternative A may be discarded. However, we cannot make such a simple choice between solutions B and C since the former is better than the latter with respect to objective Z1 but worse with respect to Z2. In general, more points (or solutions) such as B and C may be identified to define the set of all non dominated feasible solution points that form an optimal trade off curve or curve of best options.
For an unconstrained problem, further ranking of alternatives cannot be conducted without the introduction of value judgments. Specific information has to be elicited from the decision maker to determine the most preferred solution. In its most complete form such information may be summarized by a family of equi-preference curves that indicate the way in which the decision maker or society trades off one objective against the other – typical equi-preference curves are shown in Figure 1. The preferred alternative is the one that yields the greatest utility, which occurs (for continuous decision variables as shown here) at the point of tangency D of the best equi-preference curve, with the trade off curve.
Since the equi-preference curves are usually not known other practical techniques have been developed to narrow down the set of feasible choices on the trade off curve. One approach uses limits on objectives or “exclusionary screening”. For example, in the figure, the decision maker may face an upper bound on costs CMAX (i.e., a budgetary constraint). Similarly, ecological experts might set a maximum value of biodiversity loss BMAX (e.g., a level beyond which the ecosystem collapses). These two constraints define a more restricted portion of the trade off curve (darker line), thereby reducing and simplifying the choices available.
Pearce and Turner describe five main forms of multi-criteria evaluation methods: aggregation, lexicographic, graphical, consensus maximizing, and concordance[2]. Among the various types of multi-criteria analysis, the most suitable method depends upon the nature of the decision situation[3]. For instance, interactive involvement of the decision maker has proved useful for problems characterized by a large number of decision variables and complex causal interrelationships. Some objectives may be directly optimized, while others will need to meet a certain standard (e.g., level of biological oxygen demand (BOD) not below 5 milligrams per liter).
The major accomplishment of MCA models is that they allow for more accurate representation of decision problems, by accounting for several objectives. However, a key question concerns whose preferences are to be considered. The model only aids a single decision maker (or a homogeneous group). Various stakeholders will assign different priorities to the respective objectives, and it may not be possible to determine a single best solution via the multi-objective model. Also, the mathematical framework imposes constraints upon the ability to effectively represent the planning problem. Non linear, stochastic, and dynamic formulations can assist in better defining the problem but impose costs in terms of complexity in formulation and solving the model[4]. In constructing the model the analyst communicates information about the nature of the problem by specifying why factors are important and how they interact. There is value to be gained in constructing models from differing perspectives and comparing the results[5]. MCA used in conjunction with a variety of models, and effective stakeholder consultations, could help to reconcile the differences between individual versus social, and selfish versus altruistic preferences.
In addition to facilitating specific tradeoff decisions at the project level, MCA could also help in selecting strategic development paths.
Notes
Further Reading
- Munasinghe, M., 1992a. Environmental Economics and Sustainable Development, Paper presented at the UN Earth Summit, Rio de Janeiro, Environment Paper No.3, World Bank, Wash. DC, USA.
|
|
