Chapter 3
Table of contents Chapter 3
Chapter 3: Literature survey 3-2
3.1. Introduction 3-2
3.2. Project Selection Methodology 3-2
3.2.1. Benefit contribution models 3-3
3.2.2. Comparative approaches 3-4
3.2.3. Optimisation methods 3-6
3.2.4. Strategic planning methods 3-7
3.2.5. Ad Hoc methods 3-8
3.2.6. Conclusion on project selection methodologies 3-9
3.3. Conclusion of Chapter 2 3-10
Table of figures Chapter 3
Figure 21: Project selection process 3-2
Figure 22: Multiple test framework for project selection (Chapman et al, 2006) 3-7
Figure 23: Systems approach to project selection 3-9
Chapter 3: Analysis of existing theory
3.1. Introduction
The literature survey examines the subject of project selection. The literature on portfolio selection, programme selection and technology selection must still be studies and added here. The issue of R&D project which is addressed under project selection methodology in this chapter must also still be addressed.
3.2. Project Selection Methodology
The project selection process is shown in Figure 21 and consists of the soliciting of proposals, the application of a methodology utilising criteria, the output of which is a set of selected projects. This process is cyclical and is repeated each time new projects are selected.
Figure 21: Project selection process
When repeating the cycle to select new projects, the projects currently being executed also need to be taken into account (Winkovsky et al., 1980).
Project selection amounts to decision making under uncertainty as there is normally a lack of information about the future potential of research ideas (Costello, 1983). Project selection is required due to the fact that organisations do not have unlimited resources and need to make maximum effective use of available funding (Hall and Nauda, 1990). In the African scenario this is particularly important due to the limited availability of funding and other resources.
It is crucial that the correct projects are selected as choosing the correct projects can mean the difference between remaining competitive and falling behind (Martino, 1995: vii).
The failure to select the “best” project has a twofold cost. The first cost is that of the actual resources expended on failed projects. The second is the cost of the opportunities lost on projects that could have succeeded with the wasted resources (Martino, 1995, p1). Africa cannot afford to not select the “best” technologies due to the limited resources available. Not only is the funding limited, but skilled human resources are also limited. For this reason Africa cannot afford to waste resources by not selecting projects properly.
The selection of projects is a complex problem that entails multiple dimensions. The dimensions entailed can be classified as quantitative and qualitative (Cook and Seifford, 1982). Quantitative dimensions normally entail the economic measure of the project, i.e. future savings in capital, operational and maintenance costs, profits, improvement in productivity, etc.
In terms of sustainable energy projects the quantitative measures would include the cost effectiveness of the solution, the capital required for the implementation, operational and maintenance costs, and the revenue that can be collected by selling the concept elsewhere in the world.
The qualitative dimensions are those dimensions that are difficult to measure and include aspects such as political and senior management support, client support, public support, environmental impact, technical and educational relevance, degree to which project interfaces to other ongoing projects, impact on project portfolio, to name but a few (Cook and Seifford, 1982). For sustainable energy projects in Africa, the qualitative dimensions would include the potential of the project to ensure sustainable energy for the people, the potential of the project to contribute to skills development and job creation, and the potential of the project to improve the science and technology base in Africa.
In addition to the abovementioned qualitative dimensions, the most important selection criteria of course remain the likelihood of the success and the likelihood of the implementation (Cook and Seifford, 1982). This will need to be taken into account for the selection of sustainable energy projects in Africa. The ultimate goal must be to have results that can be implemented.
A summary of the main project selection methods is given in Figure 1-1. The following sections summarise the different project selection methodologies and their advantages and disadvantages.
3.2.1. Benefit contribution models
The benefit contribution models are economic models that attempt to compute the cost benefit of performing a project or attempt to quantitatively assess the financial risk of performing a project (Hall and Nauda, 1990).
Net Present Value (NPV) converts the cash flow of projects to a single value, stated in present monetary value, which makes comparisons between early and late values in the same cash flow stream possible as well as a comparison between cash flows that have different profiles of income and expenditure (Martino, 1995). In a survey by Cetron et al. (1967), nine of the methods that were examined utilised NPV. NPV allows for the comparison of project with different streams of expenses and revenues. The main problem in the utilisation of NPV is that cash flows for R&D projects are not very predictable. A further problem with NPV is that it assumes a constant discount rate over time (Martino, 1995).
The Internal Rate of Return (IRR) is the discount rate that would reduce the NPV of a cash flow profile of a project to zero. For the selection of projects, the greater the IRR, the better the project as the project will achieve payback sooner (Martino, 1995). The advantage of this method over NPV is that future interest rates need not be estimated, but just as with NPV, the future cash flows of R&D projects must be estimated.
The cost benefit model proposed by Silverman (1981) combines a scoring/economic approach for estimating the relative merits of R&D projects. The method requires the estimation of three vectors of economic and scoring values i.e. energy benefits, consumer savings and societal factors. The advantage of this method is that it focuses on managerial issues but that is to the detriment of the technical project issues that are not addressed.
As an example of a risk analysis approach, Sefair and Medaglia (2005) propose a mixed integer programming model that maximises the sum of net present values of chosen projects, while minimising the risk of the projects. The model combines the project selection and sequencing decisions while considering risk and profitability as optimising criteria. The advantage of the approach is that it takes more factors into account than the NPV approach. On the other hand, the risks of R&D projects are not always easy to quantify, especially over the longer term.
3.2.2. Comparative approaches
Comparative approaches are based on group evaluation of projects by comparing one project to another.
For ordinal ranking, each member of a committee is asked to rank a set of projects ordinally along a set of dimensions. It is then assumed that cardinal weights are assigned to each dimension which are utilized to simplify the problem into a single dimension. An index indicating the degree of agreement of the committee members is given. A constrained linear assignment model is then used to allocate the relative project priorities (Bernado, 1977 as referenced in Cook and Seifford, 1982).
The ordinal ranking model is simple and easy to use. Despite the advantage of simplicity, the disadvantages include that the model assumes that dimensions can all be collapsed through the use of a set of weights, which is equivalent to proposing the existence of a utility function. The model is also structured for small problems and will be cumbersome for over 50 projects (Cook and Seifford, 1982).
Q-sort is a structured group communication psychometric method for classifying a set of items according to the individual judgment of a group of persons selecting the projects. Each individual successively sorts items into preconceived categories. The anonymous scores are tallied and these tallies are then used as a starting point for open discussion (Souder, 1978).
This method is a valuable procedure for facilitating scientist/scientist and scientist/manager communications within project evaluation process as a clear indication of the opinions of the various group members is obtained (Souder, 1978).
Participants on a Q-sort experiment felt that the method was too imprecise to yield final decisions. They also felt that the process was highly subjective to personal preferences, ignorance and misunderstanding (Helin and Souder, 1974). Archer and Ghasemzadeh (1999) comment that the process is cumbersome as the large number of comparisons involved have to be redone if another project is introduced.
When using the pairwise comparison method, projects are compared (e.g., preference for project i against project i+1, project i against project i+2, etc) until every pairwise comparison is explored (Hall and Nauda, 1990). The most common methods for converting the comparisons into rankings are the dominance count method and the anchored scale method (Martino, 1995).
A more sophisticated approach that also utilizes pairwise comparison is discussed by Mohanty (1992). In this approach gives a final acceptability index for each project that is used to rank the set of projects.
The main advantage of pairwise comparison is that it elucidates conflicts and differential perceptions of R&D objectives. It also induces articulation of value structures and disclosures of hidden social-interpersonal conflicts (Souder, 1975).
The disadvantages are once again that the comparisons have to be redone if another project is introduced (Archer and Ghasemzadeh 1999). Martion (1990) also notes that this method can lead to a large number of ties among the projects in the middle.
The Analytic Hierarchy Process (AHP) is conducted in two stages namely hierarchic design and evaluation. (Saaty should be referenced as he is the ‘father’ of AHP) Design of the hierarchy involves structuring all the elements of the selection problem into a hierarchy. The method is based on determining weights of a set of criteria in one level of the problem hierarchy to the level just above. By repeating the process level by level, the priorities of the alternatives at the lowest levels can be determined according to their influence on the overall goal of the hierarchy (Liberatore, 1987).
The main advantage of AHP is that it allows the R&D project selection problem to be linked to the business strategic planning process (Liberatore, 1987). The disadvantages are once again that the comparisons have to be redone if another project is introduced (Archer and Ghasemzadeh, 1999).
Scoring models require individuals to specify the merit of each project proposal with respect to available and determinable criteria. The scores are then aggregated to determine an overall project rank. The highest ranking projects that can be performed within budget constraints are selected (Hall and Nauda, 1990).
Scoring models have many advantages including simplicity of use and formulation. Scoring models can also take into account both objective and judgemental data (Martino, 1995) and projects can be added and deleted without recalculating the merit of other projects (Archer and Ghasemzadeh, 1999). The value of a scoring model is however based on how the decision criteria are selected, whether these criteria are really known or based on estimates.
3.2.3. Optimisation methods
These types of models seek to optimise some objective function or functions subject to specified resource constraints. Different authors use a number of different objective functions, which are normally economically based, and different constraints to formulate the project selection problem. These methods are conceptually attractive as they optimise specific quantitative measurements of R&D performance subject to budget and organisational constraints. Surveys have however shown that these models are not very widely used (Archer and Ghasemzadeh, 1999).
Various types of optimisation models exist including integer programming, linear programming, non-linear goal programming, non-linear dynamic programming and a multiple test framework.
Integer programming consist of an optimization where the variables may only take integer values, i.e. 0,1,2,3,....
A value vl is assigned to each project l. The cost cl of funding that project is determined. The binary knapsack problem must then be solved:
Maximise
Subject to
xl = 0 or 1
where B is the available budget. xl = 1 implies that the project l is funded (Cook and Seifford, 1982).
The advantage is that this is a very simple integer programming problem to solve. The problem is that the values and costs are not always available in an objective manner and the degree of preference of one project versus another needs to be expressed. In many cases it is unrealistic (Cook and Seifford, 1982).
The other programming techniques all have similar formulas that can be solved utilising a computer programme.
The multiple test framework proposed by Chapman et al. (2006) consists of a traffic light process where individual projects are submitted to six tests, each of which has a simple traffic light outcome. If a project obtains a green light for all six measures it is accepted. A red light on any of the measures means immediate disqualification. A project with one or more orange lights is reconsidered at the next planning phase. This framework is shown in Figure 22.
Figure 22: Multiple test framework for project selection (Chapman et al, 2006)
This method allows for more criteria than purely NPV to be taken into account. For marginal and complex choices however the review process becomes a lot more difficult (Chapman et al., 2006).
3.2.4. Strategic planning methods
Various strategic planning methods are discussed in the literature. In this proposal cluster analysis, decision tree diagramming, decision process models and expert systems are considered in more detail.
Cluster analysis focuses on selecting projects that support the strategic positioning of an organisation. In essence the list of projects is taken and clustered together in a hierarchy according to their degree of similarity. A cluster or clusters of projects are then funded that support the organisational strategy (Martino, 1995: 76).