Participatory multi-criteria decision analysis: a new tool for integrated development planning

Abstract

This article proposes a new method for implementing integrated development planning as envisioned by the Local Government Municipal Systems Act of South Africa (2000). The method draws together stakeholders who fall into three broad groups: the communities who live in the municipal area, municipal officials and the elected local politicians. It is grounded in the principles of participatory action research, in which the participation of all interested and affected parties is valued. Within this participatory framework, tools of multicriteria decision analysis (MCDA) are used to support the decision-making process by structuring and quantifying difficult decisions that need to be addressed. Community development measurement scales are formulated by communities, and form the basis for evaluations of proposed projects and ongoing monitoring of these communities' progress.

1. Introduction and background

The Local Government Municipal Systems Act No. 32 of 2000 (South Africa, 2000) requires that municipalities prepare and submit for scrutiny an Integrated Development Plan (IDP), which involves the integration of three aspects of democratic governance: public participation, public administration and planning and political leadership and guidance. There are thus three broad groups of participants in the IDP: the public, the municipal officials and the elected political representatives. The methodology proposed in this article focuses on the public participation aspect of the IDP and on how the information gathered from this arena can be integrated with that from the other two participant groups of the IDP.

The IDP has three overriding objectives: to provide for poverty reduction and sustainable municipal development while maintaining a sound environment (see Guide Pack 3, South Africa, 2001). The IDP process involves specifying priorities for the municipality, which requires agreement on value judgements, informed as far as possible by pertinent information. It also requires that proposed alternative courses of action (in the form of strategies or projects) are evaluated by the participants, which involves a degree of judgement as to the efficacy of the proposed actions. It thus involves multiple (possibly conflicting) role players dealing with multiple conflicting goals. The proposed methodology outlines the use of Multi-Criteria Decision Analysis (MCDA) tools to (a) clarify the needs and aspirations of the communities, (b) evaluate a number of alternative plans to address these needs and (c) monitor progress in the communities. An IDP process culminates in the formulation of a budget for the forthcoming period. Projects and interventions which participants believe will effectively and sustainably support the municipality's developmental objectives are included in the budget. Figure 1 summarises the inputs and outputs to the process of drawing up an integrated, participatory budget, via an integrated developmental planning process. The constraints referred to are both financial and technical. While broad constraints (such as the budget available for capital projects) are communicated to all participants at the beginning of the process to contextualise their decision making, a tool is proposed wherein the constraints also become an input to the technical process of formalising a collection of possible interventions (plans) into a suggested final budget.

Figure 1: Inputs and outputs for the proposed process to compile an integrated, participatory budget

This method was developed within a participatory action research paradigm, working with the communities of the municipality of Stellenbosch during their 2000/2001 IDP cycle. The overall aim of the research was to build a framework for the continuous and participatory development, monitoring and evaluation of municipal plans, including the budget.

2. What is multicriteria decision analysis?

‘Decision making’, whether conscious or unconscious, is an ongoing human activity – it is part of everyone's daily life. The discipline of MCDA is a branch of decision analysis (or theory) which deals explicitly with decision making in the context of simultaneous, multiple goals, criteria or objectives. In reality, however, all decisions have more than one dimension or criterion to them or else there would not be much deciding to be done. What then is the difference between MCDA and other formal approaches to decision-making problems (such as, for example, mathematical optimisation)? [This is a mathematical technique for maximising/minimising a quantity subject to constraints on resources.] Consider the example of the selection of an investment portfolio where there is a desire to maximise growth, maximise dividend yield and minimise risk. A solution to this can be found by, for example, maximising growth subject to given constraints on dividends and on associated risk. Thus the problem can be made to appear as though it only has one objective (and two constraints), and a solution can generally be found for different levels of the constraints. However, clearly many potentially satisfactory solutions could be missed using this approach. An MCDA approach would focus on exploring alternative solutions to this problem by ‘juggling’ all three objectives at once and not converting two of them to constraints; hence the emphasis on the ‘multiple criteria’, as opposed to formulating the problem as one with a single criterion/objective and multiple constraints.

Roy (1999: 4) points out the dangers of what he refers to as arithmo-morphism, in which all dimensions of a problem are reduced to one single dimension, usually money, taking no account of the appropriateness of the transformations or issues of scale or other hidden assumptions of equivalence. He concludes that it was partly as a move away from this kind of reductionism that MCDA evolved as a powerful new approach in which ‘[decision-aiding]… is supported by multiple scales, which, in general, cannot, in any objective way, be reduced to, or converted in[to] a single one. In such conditions, rather than dismissing or concealing subjectivity, it is important to make an objective place for it which will be compatible with a plurality of expression’. This highlights, too, the ability of MCDA to simultaneously embrace decision criteria (or dimensions) which range from qualitative through to quantitative and also to allow for individual expressions of value judgement.

Essentially, all forms of MCDA encompass three broad actions: explicit identification of criteria, evaluation of alternatives with respect to individual criteria, and aggregation across criteria. An important and distinctive feature of MCDA is its separation of issues relating to fact (and certainty/uncertainty thereof) and value. Keeney (1992) refers to this separation as distinguishing between beliefs (which he associates with facts) and values (which relate to feelings about what is important).

What does formal MCDA do? As stated above, it makes explicit the competing nature of the different aspects of the problem at hand and also actively allows for subjective judgement. The focus then becomes one of developing insight into complex problems rather than necessarily finding the ‘right’ or ‘optimal’ solution. Very broadly speaking, MCDA is a process of structuring a complex problem in terms of criteria and alternatives. As part of the MCDA process, the set or space of possible alternate solutions is explored in a way that generally assumes there will be compromise and trade-off in a situation where the ‘ideal’ is often not attainable. Keeney & Raiffa (1976: 66) sum it up by stating: ‘Generally you simply cannot maximise several objectives simultaneously… nor can you share a pie by giving the maximum amount to each child’. Thus there is usually some notion of an ‘efficient frontier’ of alternatives that in some sense are ‘best’ for a given level of compromise.

3. Stakeholders and participants in the IDP process

The stakeholder groups involved and interested in the IDP process are broadly outlined below.

Government officials and associated experts: this group consists of municipal managers, sector heads (which could include managers from other spheres of government as well), treasury officials, and all those responsible for managing the execution of municipal functions. Also included in this group are any experts whom the municipality may choose to engage for assistance with their planning and budgetary processes. These participants are nominally responsible for implementing the budget, but they also have a responsibility for enabling the other participants to contribute meaningfully to the process. This group has the technical knowledge and know-how about the functioning of the municipality.

Councillors/politicians: these are the democratically elected representatives of the people. They hold the ultimate decision-making power as far as approving both the IDP and the budget is concerned. They also have a role to play in ensuring that the IDP is an effective and coherent tool for implementing an agreed set of policies.

Communities: this group includes residents, ratepayers, advocacy groups, civic organisations, labour organisations, employers and other interested and affected parties resident in or active in the municipal area. They may be divided into subcommunities, which may be geographically based. It should be noted that with respect to the IDP the ‘poor and other disadvantaged groups’ are specifically highlighted within the definition of a community as given in the Local Government: Municipal Systems Act No. 32 of 2000 (South Africa, 2000). As such, specific measures should be introduced to ensure that their needs and issues are captured.

Facilitators: these are ‘neutral’ person(s) (i.e. those not having allegiance to any of the participant groups) who facilitate the workshops, collate the information provided by the various participants and ensure that the process retains momentum to reach the agreed deadlines. It is the role of the facilitator to ensure that the process is as participatory as possible, and to help participants to be constantly aware of the overall aims of the IDP whilst still doing justice to their particular constituents.

District, provincial and national government representatives: these officials (usually IDP managers within the different spheres of government) are in the main recipients of information needed for their own IDP or other planning processes. However, District, Provincial and National strategic plans and objectives form an important context for the IDP process as well.

4. Participatory MCDA: An overview of the method

The proposed decision-making method consists of the following steps (as outlined in Figure 2). Technical details of each step are spelt out in Section 5 below.

Figure 2: Steps towards an integrated, participatory budget

Step 1. Gather information: collate (and share) input from the various role players in the form of workshops, surveys, technical studies, etc.

Step 2. Specify, group and prioritise needs:

2a. Draw up value trees for each community: each constituent community defines a hierarchy of criteria/issues/needs for their area or domain of interest.

2b. Define community development measurement scales. Draw up scales for each dimension of the value tree for each community. These scales may be linked to outcome measures (indicators) and specific means of collecting data to inform them.

2c. Use weighting techniques to reach agreement on priorities for the issues making up the value trees.

Step 3. Strategise, evaluate and plan:

3a. Formulate objectives and strategies. Specific alternative solutions to the identified needs or issues are generated. These may be in the form of broader strategies (ultimately groups of projects) or specific projects.

3b. Evaluate the alternatives by scoring on a number of criteria. The criteria used to score the identified alternatives should include impact (in terms of the community development measurement scales devised in 2), cost and sustainability of the solutions.

3c. Explore the set of alternatives to find the most efficient budget (collection of alternative projects or strategies). Search for the combination of projects (or strategies) that will provide maximum impact for minimum cost and that is most sustainable. Note that this step is not specifically included in the participatory process as laid out in the IDP documentation; however, because there are technical tools which could greatly assist this process, some suggestions are put forward in this article with the idea of exploring these further in future work.

5. Steps of participatory MCDA

5.1 Gather information (Step 1)

All role players need to be involved in this step which is vital to the integrity and transparency of the process. There are likely to be information gaps (in terms of what it would be desirable to know before embarking on the process) and all parties need to have access to that information which is available, and be made aware of what remains unknown at this stage. This is an opportunity for all players to inform each other from their strengths. For example, communities bring their expertise and experience about issues in their areas and government sector officials contribute their knowledge of technical solutions and financial constraints.

• (i) Municipal officials need to present to the public the status quo in terms of current community development (levels of housing, service provision, gaps in delivery) as well as outstanding issues to do with maintenance, challenges and opportunities facing the municipality, fiscal and statutory requirements and limitations, and available resources. This information is generally held by (or accessible to) the municipal officials and is vital for the communities to provide a context within which to table their needs. Where possible, information should be collected at the same scale for all areas and where not possible this must be made explicit. Problems of data availability or of scale frequently arise because of the involvement of a number of different authorities or service providers. As examples of such problems, electricity may be provided to one part of a municipal area by the municipality itself and to another by the electricity supply company, Eskom; or some areas may have their health services provided by the local municipality and others by the district municipality, and information about health issues may well be collected by these two bodies using different area boundaries.

  If there are gaps in required information at this stage or if the data is not current (for example if census data is used) or reliable, then this must be noted and addressed within the scope of the IDP.

• (ii) Community representatives are given a chance to table their needs, issues and aspirations. At this stage these may be reflected as needs, issues, problems, comments, policies, values, projects or strategies. This information may be collected via (a mixture of) workshops, sample surveys, focus groups or targeted interviews. It comes from the community's perceptions and actual experience of their own levels of development and may well be supported by objective information where available (specifically that presented in Step 1).

  If there are gaps in this information they are likely to be due to under-representation of particular groups, so the design of this stage should take into account the need to work through established community structures such as ward committees, and also to be alert for ‘missing voices’ of marginalised and less well-organised members of the community.

• (iii) Politicians need to take this opportunity to clarify how the IDP is meant to be used as a tool to implement a certain political vision and to ensure that particularly important groupings and issues (such as poverty and AIDS) are not allowed to fall off the agenda. The facilitators may need to take up this role if it is not forthcoming from the politicians (for example, highlight the need to analyse gender, poverty and HIV/AIDS issues or note that the information on these issues is insufficient).

• (iv) Representatives from other spheres of government need to be kept informed of the process but may not be directly involved in this step, except to clarify their roles as service providers (for example library services) where appropriate.

5.2 Specify, group and prioritise needs (Step 2)

5.2.1 Draw up value trees for each community (Step 2a)

A value tree is a structured (usually hierarchical) set of criteria or issues. The usefulness of conceptualising such hierarchical arrangements of objectives/criteria, where higher order goals are broken down into lower order or subgoals, has been well established by, among others, Keeney (1992).

This step involves structuring the needs of the community into a value tree that will form the basis for defining needs and goals and measuring progress in the community. It can be useful to structure needs in a hierarchy, in order to begin to allocate resources more rationally and purposively, and monitor consumption. Depending on available capacity to engage meaningfully with communities, a municipality may need to draw up a number of value trees, i.e. one for each identified constituent subcommunity.

Although the IDP process encourages the integration of issues, it is important first to develop a representation of what the separate issues are and how they may be broken down into subissues before looking at the effect of, say, a strategy across different issues. The value tree displays (catalogues) the issues for a particular community or area. An example of a simple community value tree is given in Figure 3.

Figure 3: A simplified example of a hierarchy of issues for a hypothetical community

The role players involved in this step are:

• The community representatives: each distinctive community structures their needs as far as possible, making use of a generic template within which to record their issues as a starting point. Each community has an opportunity to shape the template to accommodate their own distinctive needs and issues. In general, however, communities will record their issues under a set of headings (referred to as clusters) which capture the higher order objectives of the municipality. Subclusters may differ from community to community. The information used at this stage is that tabled by the communities in Step 1.

• The facilitator(s): the facilitator chairs and coordinates the meetings or workshops needed for this step of the process and focuses the participants to find the preferred way to providea basis for comparing needs across areas, and a basis for establishing priority needs for the municipality as a whole as well as for each area.

• The politicians: the politicians clarify the high level objectives for the municipality within the context of national- and provincial-level imperatives.

5.2.2 Define community development measurement scales (Step 2b)

In order to be able to measure progress in the various dimensions (clusters and subclusters) that are ascribed to community wellbeing it is important to define some endpoints and/or benchmarks. For each dimension of the value tree, a qualitative scale should be devised which defines:

• (i) the goal, or a description of the best outcome for this community

• (ii) a description of the worst outcome for this community and

• (iii) one or two intermediate points between the descriptions given in (i) and (ii).

Note that this descriptive scale is similar to Drewnowski's ‘critical points’ (1974), in which he divides an indicator into a number of benchmarks or subranges (he suggests four). The more points that are defined on the scale, and the fuller these descriptions are, the more useful the scale will be in helping a community measure the impact of an intervention/project/budget or the performance of a given period of municipal administration. Some examples of hypothetical best, worst and intermediate outcomes for clusters of the community value tree in Figure 3 are given in Table 1. Also tabled are possible indicators that might arise from the dialogue around defining these points of the scale.

Table 1: Hypothetical community development measurement scales

Measure progress	Housing & land	Roads transport	Recreational facilities	Jobs	Health education
100	Best outcome: All people have access to adequate, weatherproof, safe houses that are effectively financed, close to job opportunities and amenities and linked to affordable services	Best outcome: There is a safe, effective and efficient system of roads and public transport which links people's homes to their places of work, schools, recreation and other amenities	Best outcome: The community have adequate spaces and structures in which to congregate, socialise and play. There are effective facilities for the elderly, the disabled, the youth and children	Best outcome: All who are able and available to work have access to employment or an acceptable means of generating an income. Youth in particular have access to employment	Best outcome: All people have access to clear and accessible information about healthy lifestyles and safe practices. There are effective campaigns to encourage adherence to these principles.
50	Roughly three-quarters of the community are adequately housed	Roughly three-quarters of the public have access to some form of public transport that is affordable to them. There is limited provision for the transport of schoolchildren	There are some recreational facilities for the community but they are either too few or inappropriate for the above designated groups	Around 50% of the community (or the youth) are unemployed OR less than one quarter of the community derives income from drugs/crime/prostitution	There is effective health educational material but it is estimated to reach and educate around 50% of the population only
0	Worst outcome: More than half the population does not have access to adequate housing	Worst outcome: Roads are inadequate and unsafe. Public transport is unaffordable or unavailable to most. Pedestrians, especially children, are at risk on the roads	Worst outcome: There are either no recreational facilities for the community or very inadequate ones	Worst outcome: More than three-quarters of the community is unemployed with no prospects OR at least one third of the community derives income from drugs/crime/prostitution	Worst outcome: There is no information on healthy lifestyles and practices, or it is ineffective or inaccessible
Possible indicators:	Percentage population housed Average distance to work Proportion with access to services Proportion of houses with defects	Road/pedestrian fatalities Average time/cost to reach work Average time/cost to reach school	Community facilities per unit of population Average distance to nearest sportsfield/hall/youth centre	Unemployment rate Youth unemployment rate Female unemployment rate Crime statistics	Community perceptions of Aids Compliance rates (TB medication) Alcoholism Drug dependencies Teenage pregnancies Foetal Alcohol Syndrome (FAS) Distance to nearest clinic

As a guideline to drawing up these descriptive points, it is suggested that the community consider all the possible dimensions of the issues raised under a particular cluster (for example, the issue of roads may have a safety aspect, a transport aspect, an employment opportunity aspect, etc.). Target groups within the community that particularly need to be monitored (the elderly, the disabled, children, etc.) should be specified. It is also important to include relevant associated or covariate issues (effect of housing on health, effect of education on social problems, etc.). A covariate can be thought of as an issue which ‘interferes’ with another variable (either positively or negatively). For example, sturdy, weatherproof housing is likely to have a positive effect on the health of the community.

Ultimately it would be beneficial to describe three to five points on the community development measurement scale, and nominally assign the best outcome a score of 100 and the worst outcome a score of 0. The point midway between these two would then score 50. A similar process could be used to ascertain the points that correspond to the 25 and 75 scores. This scale creates a very basic and essentially qualitative tool for a community to measure progress in a particular issue (or dimension), allowing for interpolation between the benchmarked points where necessary. The process of defining points on the qualitative scales and the ensuing debate very often leads naturally to the identification of objective and directly measurable indicators of progress through the nature of the descriptions used for the benchmarks (examples as given in Table 1). The role players involved in this step are the following:

• The community representatives,who cooperatively write scenarios that describe a lack of development, partial development and full development in each of the dimensions identified in Step 2a. These are informed by the information tabled in Step 1.

• The facilitator(s), who assist by clarifying the aims of this exercise (as laid out above) and may also provide examples as a starting point from which the community can work.

• The politicians, who receive the information from Step 2b.

• The municipal officials, who receive the information from Step 2b.

5.2.3 Use weighting techniques to reach agreement on priorities for the issues making up the value trees (Step 2c)

One of the underlying aims of following a quantitative approach to monitoring the IDP process is to be able to measure the impact of a budget on the issues identified as important by communities. To measure overall impact requires two separate measurement processes:

• (i) measuring the relative importance (weight) of the identified issues and

• (ii) measuring (scoring) the impact of a proposed intervention (budgeted programme or project) on the identified issue.

An overall ‘value’ for a budget or budgetary intervention can be assigned using a value function. This simple approach essentially evaluates a weighted average of the impact of an alternative (intervention), taken over all the different community issues. There are naturally more sophisticated ways of combining scores and weights (see Belton & Stewart, 2002, for a comprehensive review of such methods) but an advantage of this approach is that it is transparent, easy to explain, logically defensible, and does not involve complicated or suspicious looking mathematical formulae.

There are various ways to elicit the weight parameters associated with the various criteria. In order to use a weighted average as a means of obtaining an overall score for an intervention/budget, it is important that these weights are measured on a ratio scale. Ratio scales have the property that ratio relationships between two numbers are meaningful but differences are not. A so-called ‘swing weighting’ approach (as proposed by von Winterfeldt & Edwards, 1986) is an appropriate device for deriving weights for an additive value function such as is being proposed here, as it also preserves their algebraic properties as scaling constants. Thus not only must the weights be measured on a ratio scale, they must also reflect the relative importance of a swing from worst case to best-case scenarios. In other words, one needs to explore concepts such as ‘to bring about the envisaged change in A is the most important’, ‘the change in B is only half as important as that in A’, and ‘the importance of effecting a change in C lies somewhere in between these two’. A way to accomplish this task is to ask community representatives to imagine ‘filling up beakers’ to represent the importance levels assigned to bringing about a change from worst to best in the different criteria, as shown in Figure 4.

Figure 4: A pictorial representation of the swing weights allocated to community issues

The issues are assigned a weight equal to the amount of liquid added to each beaker, so the weight of A = 100, B = 50, and C = 75. [Note that, technically speaking, respondents are asked to compare the heights of the ‘liquid’ in each beaker and not the volume, i.e. the illustration shows a cross-section of the beaker so that filling it to the 50 mark means it is half full, and filling it to the 75 mark means it is three-quarters full, etc.] The representative groups adjust the weights by iteratively comparing the levels of liquid in each pair of beakers until consensus is reached as to the relative importance of all issues. This process involves, in some cases, many hours of discussion and ‘lobbying’ before consensus can be reached, and consensus may only be possible if there is not extreme polarisation in the group to start with. In some instances it may be necessary to split an issue into component subissues before agreement can be reached. In other words, disagreement on weights may represent fundamental differences in the way in which a cluster is viewed. (For example the word ‘Health’ may represent different things to different people, so breaking the concept down into ‘mental’; ‘physical’ and ‘emotional’ health may clarify the picture and allow weights to be allocated more unanimously).

The weighting procedure is essentially two-stage:

• (i) Determine internal weights: work separately within each cluster to determine the relative weights of each of the issues at the given time and under the prevailing circumstances for the community. Participants are asked to consider the defined descriptive scale (or more specifically the best and worst case scenarios) for each issue. They are asked to consider the situation where the community is at the worst-case scenario for each of the criteria. They are then asked to consider the situation where they could change one of the criteria to the best-case scenario as defined for that particular criterion. They are asked to select which criterion that would be (or criteria if they choose more than one criterion that they most wanted to swing from worst to best case). Participants are given a visual analogue tool in the form of filling up beakers to specify the relative weights. Thus, within each cluster, the issue which participants rate as the most important or pressing (to swing from worst to best) is assigned a weight of 100 (a ‘full beaker’) and all other issues are given a weight relative to this maximum value of 100. This ‘swing weighting’ approach effectively creates an equivalence scale between the units of measurement of the different issues.

• (ii) Determine the relative priorities of those issues that score 100 in each of the major clusters (cross-cluster weighting). This is done by repeating the procedure in (a) for the group of criteria consisting of those scored as 100 in each cluster of subcriteria. The weights are then adjusted algebraically to get an overall relative weighting for each issue and subissue.

The above steps ensure that all major clusters are given attention, as against an approach of directly ranking all issues simultaneously, which usually leads to issues such as roads and traffic falling off the list at a very early stage.

An example of priority weights for the issues identified in Figure 5

Figure 5: Prioritised community value tree

Suppose that within the cluster of Social Development, the particular community identify that a swing from worst to best for ‘Recreation’ would be more desirable and pressing than the same swing for ‘Welfare’, then Recreation would be assigned a weight of 100. If the swing for ‘Welfare’ is regarded by the community as roughly half as important as that of ‘Recreation’ then they might assign ‘Welfare’ a weight of 50. Continuing in this vein, the internal weights of each of the main clusters might look as follows: Social development: Recreation 100, Welfare 50; Economic development: Jobs 100, Training 30; Health: Clinics: 100, Health Education 80; Infrastructure: Roads and transport 100, Basic services 100; Natural and built environment: no subclusters; Housing and land: no subclusters.

Note that the above weights do not represent some intrinsic worth or importance level of the conceptual issue itself but rather reflect the relative magnitude and depth of the problems currently facing the community in each of the subclusters. It is important that participants understand this distinction and do not fall into the trap of debating the relative importance of, for example, Health versus Education. Such discussion is fruitless and misleading.

At the second stage of weighting the subclusters which were assigned a score of 100, and the main clusters which were not subdivided into subclusters, are compared against each other. Suppose the participants need to consider the following: Recreation, Jobs, Clinics, Roads and transport, Basic services, Natural and built Environment, and Housing and land. Of these issues, suppose that the community indicate that, faced with such a choice, the issue which they would choose to swing from worst to best would be Housing and land. This cluster is then assigned a weight of 100. The second most important swing is identified to be for Jobs; this swing is indicated to be almost as important as that for Housing and land, and thus assigned a weight of 90. Recreation follows with a relative weight of 85, Clinics with 70, Roads and transport and Basic services equally at 65 and Natural and built environment at 50. This would allow, by simple algebraic manipulation (in a manner which faithfully preserves the priorities specified by the participants), the conclusion that the priority weights of each of the subclusters are as shown in Figure 5.

The figures in the bubbles attached to each subcluster in Figure 5 reflect a priority weight or assessment, such that 100 represents maximum priority and 0 represents minimum or no priority. Note that it is important that a skilled facilitator is used to guide this process, as studies have shown that people may find it relatively easy to ‘assign weights’ to issues but might actually fail to adequately take into account the context of the decision problem and especially the range of alternative options available (see Mousseau, 1992; von Nitzsch & Weber, 1993). They need to be constantly directed to focus on the swings so that the weights do not represent some nebulous intrinsic level of importance but are algebraically appropriate as ‘trade-off multipliers’ in the additive value model.

The role players involved in this step are:

• The community representatives, who may need to consult more widely within their communities to get a consensus view of their input at this stage.

• The facilitator(s), who ensure that the participants are supported and assisted to follow an appropriate weighting procedure. They also need to indicate the consequences of, for example, allocating high weights to all issues (in which case resources become very thinly spread). The facilitator should reflect back the final weights to the participants and allow opportunity for adjustments to be made to them until the community are happy that the weights give a fair representation of their current priority issues.

• The municipal officials, who receive the information on priority issues from this step, and who may need to clarify points of understanding and also to begin to compile information on what proposed interventions are already approved, on budget, or in progress.

• The politicians, who receive the information from Step 2c.

5.3 Strategise, evaluate and plan (Step 3)

5.3.1 Formulate objectives and strategies (Step 3a)

The IDP is intended to closely inform and link to the budget which consists, among other things, of specifying a number of budgetary programmes or projects (aimed at addressing the identified priority issues). The number of interventions or projects specified can, depending on the size of the municipality, be very large. For a category B municipality (i.e. smaller than a metropolitan municipality and falling under a district municipality) this could range from 50 to 500 or more projects.

The needs of the community are likely to be specified in terms of specific projects (e.g. build a road from a to b, provide a clinic at y, etc.). For each major group of issues (cluster) it is suggested (by the IDP Guide Packs, South Africa, 2001) that the priority themes be formulated into a number of key objectives and that strategies be formulated around addressing these objectives. These strategies will ultimately be translated into a number of specific projects, possibly coinciding with the suggestions raised in the initial round of workshops. However, the step of strategising around priority themes forces the participants to rethink the issues and possibly explore alternative solutions before decidi ng on a chosen route.

Participants in this step, probably working jointly in task groups, would be:

• Municipal officials: before the task groups convene, the officials need to ensure that all of their needs and requirements (usually relating to the technical demands within each sector) are tabled alongside those identified by the communities. It may also be necessary to bring in financial (municipal) expertise so that the different routes to meeting identified needs can be explored. Rudimentary costing of strategies needs to be provided by municipal officials at this stage of the process.

• Community representatives: the community representatives need to work with the officials to identify strategies that may help address the priority needs identified by communities.

• The information that the participants use in this step of the process consists of the priority needs and the specific issues tabled in Step 1 by communities and augmented by the officials.

5.3.2 Evaluate the alternatives by scoring on a number of criteria (Step 3b)

Each option (programme or strategy) is scored with respect to a number of criteria that include impact, technical need and sustainability of the proposed interventions. There is an advantage in working with strategies rather than projects at this stage in that one is able to conceptualise a budget as a combination of strategies rather than a combination of projects. Apart from encouraging a joint and planned vision of the way forward, this is also much more tractable numerically in terms of sheer numbers of options and combinations to be evaluated. Note that at this stage the officials are asked to provide ballpark estimates of the cost of proposed alternatives. This information is kept separate at this stage and the participants are not asked explicitly to consider cost when scoring the alternatives. The reason for this is that in practice the municipality does not necessarily operate from a fixed budget. Expenditure can be financed either from the internal capital budget of the municipality, which is relatively fixed, or from ad hoc sources of funding, which are not fixed and need to be applied for and motivated. In practice, officials can shift proposed expenditure from their capital to their ad hoc budgets and so free up more available budget on the capital budget. In addition, part of the rationale behind the IDP is to encourage participants to source creative ways of financing options which have a high degree of impact on community needs (South Africa, 2001). Thus the approach here is to leave cost as a separate criterion in the process until the final stage of budgeting when it must be reconciled with impact (benefits).

In order to choose between strategies it is necessary to devise a consistent and transparent scoring system to measure the attractiveness of various strategies that can be adopted. Because the combinations of possible strategies that can be adopted to form a budget is very large, a simple scoring procedure is necessary to initially screen all the proposed strategies for combinations which might be most beneficial across all areas. Scoring can take into account various aspects or dimensions of a decision situation, the most obvious usually being benefit and cost. In the case of the IDP it is important that sustainability of solutions be considered as well. The community development measurement scales for each main cluster of issues can be used to measure the impact of a proposed strategy in each area.

A scoring system should be as simple as possible while still being transparent and consistent. If using an ordinal scale, it is important to define the meaning of different points on this scale to ensure that the scores are consistently applied. By way of example, the scoring systems shown in Tables 2 to 4 show possible ways of scoring different alternative strategies. These were drawn up by the participants in the 2001/2002 Stellenbosch Municipality IDP process. One could expect that different scales would emerge from different participant groups. Note that all three scoring instruments described below allow for interpolation between the defined points.

• (i) Impact scoring system

  This scoring system (Table 2) refers to the community development measurement scales defined in Table 1 for each subcluster of issues, to evaluate the impact that the proposed strategy can reasonably be expected to have. The numbers are used as guidelines for what constitutes a ‘low’, ‘moderate’ or ‘high’ impact, and one can interpolate between the scores shown. Different definitions would need to be agreed on for different development contexts.

  As an example, consider the measurement scale for Housing & land shown in Figure 6. A project or strategy that can be expected to house 10 per cent more people than are already housed, or to move the community up 10 percentage points on the defined community development measurement scale for the issue, could thus be scored as a 4 for impact.

• (ii) Sustainability scoring system

  This scale measures whether the impact of the proposed strategy is sustainable in terms of money and in terms of possible negative effects on people or their environment (see Table 3). The question to be asked is: ‘Is this strategy/project sustainable?’

• (iii) Technical urgency/necessity scoring system

  This scoring dimension captures the impact the municipal officials perceive that the proposed strategy will have on their ability to perform their function(s) (see Table 4).

Figure 6: Measuring impact on housing

Table 2: Scoring system for impact on community needs

Impact	Score	Definition
No impact on the issue	0	No discernible impact on the scale defined in Table 1
Low impact	1–2	An increase of 5–7 per cent points on the scale
Moderate impact	4–6	An increase of 10–15 per cent points on the scale
High impact	9–10	An increase of at least 20 per cent points on the scale

Table 3: Scoring instrument to measure sustainability

Sustainable?	Score	Definition
Not sustainable	0	Unaffordable, extremely negative to people and environment
No, but …	1–2	… can be sustained if substantial capital/resources can be made available from future budgets or if strongly negative effects on people and/or environment could be tolerated/counteracted
Sustainable, if …	4–6	… fairly moderate amounts of capital/resources are made available from future budgets or if fairly moderately negative impacts on people/environment can be tolerated/counteracted
Yes, definitely sustainable	9–10	Proposal is financially, socially and environmentally sustainable

Table 4: Scoring instrument for technical urgency/necessity

Urgency/necessity	Score	Clarification
Not urgent/necessary	0
Nice to have but not explicitly required	1–5
Required in order to reach objectives of job/service	6–8
Must be done now or service/department will collapse	9
Legally prescribed (no choice)	10	Quote relevant Act

An example of scoring alternative proposals

As a simplified example of this weighting and scoring approach to budgeting decisions, consider the case where the community has represented their priority issues as in Figure 5. Suppose that, inter alia, the following alternative proposals have been put forward for consideration in the forthcoming budget:

Plan A: build houses with associated basic services for 10 per cent of the community.

Plan B: provide for an educational health programme and build two new clinics.

Plan C: build a community development centre providing facilities for trading and small enterprises as well as assistance with small business development.

Plan A: the proposed housing project would score a 4 for ‘Impact on community needs’. In terms of ‘Sustainability’, there is some uncertainty about costs of future services being met, and the proposal scores a 5. For ‘Technical urgency’ the plan receives a score of 7 (required in order to reach the objectives of the Housing Department). The overall score for Plan A is found by combining the three scores. (Suppose that in this instance the participants have decided to weight all three dimensions of scoring equally, although this need not be so.) The overall score is thus 16/3 = 5,33. This quantity is then multiplied (weighted) by the priority weight of the Housing and land cluster to give a weighted score of 5,33 × (100/651)=0,82. [Note that the priority score is expressed here as a percentage of the sum of the weights (necessary for the weighted average manipulation).]

Plan B: the Health package proposed is assessed by the community to have a moderate to high impact on the community and is scored at 8 for ‘Impact’. The ‘Sustainability’ score is judged to be a 6. For ‘Technical urgency’ the officials deem that the proposal is long overdue and award it a score of 8. The overall score for plan B is then (22/3)=7,33. The weighted score for plan B is then 7,33 × (70/651)=0,79.

Plan C: the community assign an ‘Impact’ score of 7 to the proposed plan; a score of 8 for ‘Sustainability’ and a score of 8 for ‘Technical urgency’. The overall score for plan C is then (23/3)=7,67. The weighted score for plan C is 7,67 × (90/651)=1,01.

The maximum score for any proposal would be 10 × (100/651)=1,54. Thus all proposals could. be expressed relative to this number and expressed on a scale of, for instance, 0–10: Plan A: 5,32; Plan B: 5,13; Plan C: 6,56. This approach leads to an ordered list of proposed budgetary interventions and this information can now form the basis on which the actual budget is constructed. The information can, in the simplest possible approach, be used to allocate the budget to the highest scoring proposals. However, using the weighted scores as input we can explore more interesting questions around the combinations of plans which would give the most effective budget (taking into account both cost and effectiveness). These possibilities are explored in Step 3c.

The participants involved in Step 3b are:

• Municipal officials, who need to score the proposed interventions from a technical point of view (as viewed from the perspective of their department and their job descriptions).

• Community representatives, who score the projects from the perspective of impact on needs.

Note that the above two groups need to be involved in scoring the projects in terms of their sustainability. The following two groups are recipients of this information, which is also advertised for public comment:

• The politicians.

• Representatives from other spheres of government.

5.3.3 Explore the set of alternatives to find the most efficient budget (Step 3c)

The IDP Guide Packs describe a process in which the community are involved in identifying and prioritising their issues, in cooperation with the officials and politicians (South Africa, 2001). They are also then involved in developing strategies to address these issues and articulating a number of consequent projects of these strategies. However, there is no recognition in the Guide Packs, or other documentation supporting the IDP process, of the complexity of subsequently combining a number of (prioritised) projects into an effective budget. This task is left to the officials, who make a recommendation to the Council, who then have the ultimate responsibility for approving the budget. This comes about after many hours of ‘horse trading’ between officials, negotiating about projects and departmental budgets in an effort to squeeze the needs of each department into the overall budget. This process could be supported by technical tools (briefly outlined below) and made more transparent and participatory, while still allowing the Council the final say.

The aim of Step 3c is to finally compose the budget by selecting a ‘good’ combination (portfolio) of projects or strategies. Tools such as integer programming could be used to select a set of possible budgets that maximises the impact on the expressed community needs, subject to given budgetary and other constraints, and which reflects different weights or priorities. [Integer programming is a special form of linear programming, the branch of mathematics concerned with maximising/minimising a linear function subject to a number of constraints.] The outputs of the community weightings and scorings would be used as direct input to an integer programming exercise. Such an approach to the problem would be formulated as follows:

Objective to be maximised: the value (or benefit) derived from the budget (in terms of criteria such as impact on the community, sustainability and ability to fulfil the functions and objectives of the municipality).

Constraints:

• budget constraints

• (optionally) a ‘fair and reasonable’ distribution of resources across communities

• (optionally) given guidelines about the distribution of resources across sectors and

• other legal obligations of the municipality.

The solution to the problem will specify the strategies/programmes to include in the budget so as to maximise the above objective, given a particular set of weights.

As the councillors are the final decision makers in the IDP/budgeting approval process, a number of different budget options could be presented to them, each corresponding to different weights (and/or constraints). The weights could be chosen to reflect (as a starting point):

• (i) the various community priorities and

• (ii) a Council view of priority issues

so that it will be possible to explore how different communities are affected by different budget allocations.

Step 3c is essentially a technical step and the participants would be:

• the facilitator (or person with expertise in the programming or evaluation tool), working with

• the municipal officials, to establish the constraints that must be applied in order to produce a realistic and effective budget and

• the Council, as the final decision makers.

6. Discussion

The proposed method is presented as a means of collating and synthesising judgements from a range of different sources into a coherent framework to inform the process of drawing up a municipal budget via a process of integrated developmental planning. Compared to other planning tools such as the LOGFRAME approach or Logic Models (see, for example Cummings, 1997), it provides an explicit basis for choosing between competing options, as is required by the IDP process. It is participatory in nature (and also a requirement of IDPs) but unlike other participatory methodologies such as Participatory Rural Appraisal (PRA), it provides a rigorous, time-bound sequence of steps which is designed to culminate in the acquisition of the necessary budgetary and planning information, with all the implied choices and judgements that this process entails. Participatory MCDA makes explicit room for information relating to values as distinct from information related to facts (about development). This distinction is potentially useful in itself, and may help resolve (or focus) differences of opinion. In common with other participatory approaches, the approach proposed in this article may be bedevilled by difficulties in achieving a consensus ‘community view’ as well as the problem of the ever-shifting face of the community as different voices come to the fore and move on. Like all participatory methods it will require skilful and considered effort to ensure that dominant voices do not manipulate the situation to maintain unequal power relations. These are potential downfalls of participatory approaches in general and not just participatory MCDA. However, it is precisely within the context of these difficulties which beset participatory planning that this approach aims to structure the decision-making process, so that the basis for various choices and proposals can be made explicit.

Participatory MCDA requires considerable time and energy to implement, although once one cycle of the process has been completed, subsequent rounds may build on existing information and so the process may evolve to a point where it is less demanding to implement. A potential downfall of the approach is that it requires a fairly high level of support in terms of the provision of relevant information about the state of development in the municipality and technical expertise to develop community assessment scales, etc. However, essentially it is the requirement for effective participation which places the burden on municipalities in terms of time and other resources. A process of effective communication with a wider range of stakeholders than the immediate municipal officials needs to be supported through additional resources. What is proposed in this methodology is a means to structure that process of interaction in a way that helps municipalities meet their legal obligations for effective participation. The government, by providing Community Development Workers to under-resourced communities, has already begun to acknowledge the need for further resources to support this process. It is possible that this approach could provide the framework within which these workers could help communities participate more fully in municipal planning processes.

This method separates out the technical aspects of municipal service delivery and provides a platform for communities to render their input alongside that of technical experts, on issues in which they have a stake and could be regarded as ‘expert’. The separation of the choice problem into component parts – community values, perceptions of need, technical aspects of proposed interventions, assessments of impact on perceived needs, and an explicit focus on financial, social and environmental sustainability – allows community members to participate in planning even though they may not fully understand the technical dimensions of the issues at stake.

The scoring mechanism proposed in participatory MCDA is not the conventional value function approach of scoring each project with respect to each criterion as, with a conservative estimate of some 15 criteria (for each of the constituent areas or wards making up a municipality), and some several hundred projects to evaluate, this would be a very demanding task. The proposed method is to score each project on just three criteria: in this case, impact, sustainability and technical urgency. This in effect proposes a value tree where each subcriterion has three associated sub-subcriteria (impact, sustainability and technical urgency), and all projects score zero in these dimensions except for the criterion in which they originate. This provides for a quick screening of a large number of projects, although unfortunately it does not allow for the usual MCDA cross-dimensional effect of, for example, a Housing project scoring well in the Health dimension. The tools suggested in Step 3c provide an opportunity to explore portfolios of projects that have a synergy across the different dimensions or criteria.

This article presents a model that aims to integrate the contributions of multiple stakeholders to a joint process of developmental planning for a municipality. The feasibility of the envisaged process and the scope for consensus being reached within the required time frames of the IDP process still need to be explored, in particular for less developed and resourced municipalities. Issues such as the practical feasibility of defining common and agreed scales of evaluation (for example, of impact) for a municipality would need to be ascertained.

There is the danger that this process could raise unrealistic expectations through encouraging participants and, in particular, communities to focus on long-term community goals and to explore possible projects which could help to achieve these goals. Despite the fact that the process explicitly allows for a ‘problem contextualisation’ phase at the beginning of the process, which should clarify the scope of resources available for development, the danger remains that participants will be disappointed by the constrained pace of improvement. These feelings need to be anticipated and managed, particularly in the present environment where the emphasis and capacity of municipalities remains on basic service delivery in contrast to community perceptions of need, which may well highlight the lack of employment opportunities.

The current form of the method proposed in this article does not explicitly model or cater for uncertainty about outcomes. Thus evaluations of impact or cost are made as if there was certainty about the outcome of a project, whereas they are in effect predictions of future impact or cost and may be subject to error or uncertainty. This is an area for future development of the proposed method, bearing in mind the need to keep the computations as simple and transparent as possible.

7. Conclusions

A method for linking members of the community to the planning and budgeting process of the municipality has been proposed in this article. Although the primary focus of this approach is on community participation, important interfaces and linkages are made with other key participant groups such as municipal officials and, to a lesser extent, the elected political representatives. The tools used to quantify subjective and objective inputs and outputs to this process assist by:

• making the subjective information more transparent

• combining information in a logical, rational and justifiable manner and

• providing a valid platform for making comparisons across the various dimensions of the decision-making problem (i.e. the issues that make up the ‘quality of life’ of a community). Specific tools have been proposed, such as scoring systems and weighting techniques which can be varied to suit the needs or tastes of different communities. The specific tools used are not as important as the coherence of the underlying process.

Important to the process is the preparatory step to build decision-making on a solid platform of information. When that information is lacking it is important to create an awareness that this information is in fact a community need. This focus on reliable data/information is further underlined as the community define scales which describe ‘states of development’ of their community which can be linked, where applicable, to verifiable or ‘objective’ indicators. Once there is an identified need to monitor a certain aspect of development in a community, it raises the question of where the information is to come from, whether available information is accurate and reliable, and whether resources exist to initiate data collection. The community are involved in defining, monitoring and evaluating their own outcomes.

The method proposed needs ultimately to be integrated with processes which are centred within the municipal authority and which feed into the more technical requirements of the IDP and budgeting. In contrast to the outputs of many unstructured exercises in community participation, which result in information overload (and consequent underuse), it is claimed that participatory MCDA yields a synthesised, defensible and usable set of information which is both quantitative and qualitative.

Additional information

Notes on contributors

Leanne Scott

Senior Lecturer, Department of Statistical Sciences, University of Cape Town. This article is based on work which was funded by South Africa Netherlands Programme on Alternatives in Development (SANPAD).