Analytical Hierarchy Process - AHP - Research Paper Example

Analytical Hierarchy Process Introduction The construction industry is one of the backbonesof the economy of many countries. In this respect, the importance of a healthy construction industry in such countries is beyond doubt. Industry-wide studies on the performance of the construction industry (NWPC, 1; Latham, 94; Egan, 34) have pointed to some key improvement areas, one of which is the use of an appropriate procurement method. Procurement is critical as it determines the overall framework embracing the structure of responsibilities and authorities for participants within the building process. Therefore it is a key factor contributing to project success. Nevertheless, procurement strategy selection decisions have been mostly judgmental, and hence subject to biases of the decision-maker. In this respect, the need to have a more objective and systematic selection model is self-evident. Developing a model for procurement selection is therefore of strategic importance and several leading research works have been reported (Masterman and Gameson, 84; Skitmore and Marsden, 121; Love et al., 221). This paper reports a study conducted in Hong Kong on the development of a procurement selection model based on multi-attribute utility technology (MAUT). The robustness of the model is enhanced with the use of the analytical hierarchy process (AHP) to determine the importance weightings for the selection criteria. The proposed method seeks to improve the objectivity of the procurement selection process by utilizing the expertise of industrial experts to determine the utility factors. The analytical hierarchy process is used also to determine the importance weightings of the selection criteria. Research Design When construction clients face the problem of selecting a procurement method for a project, the clients’ previous experience plays an influential role. As well as the degree of subjectivity, the decision is complicated by the multitude of decision criteria and options available. The nature of the problem fits nicely with multi-attribute utility technology (Edwards, 3; Green, 1992). MAUT is a methodology that can be used as a tool for measuring objectivity in an otherwise subjective area of management (Fellows et al., 27). MAUT is considered to be appropriate as it enables the integration of both the effects of the priority rating and the utilities derived from the respective procurement options. The approach involves first assigning a rating to each of the selection criteria. The ratings are then applied to the utilities derived from a particular procurement method against each of the selection criteria. By indicating the relative utility of each procurement strategy against each criterion on a numerical scale, it is possible to obtain a set of utility factors. The utility factors are in effect relative measurements of the suitability of a certain procurement strategy for a given criterion. Selection of a procurement method is subsequently based on the procurement option that provides the highest utility. The procedures used in conducting this study are summarized as follows. (1) Determination of selection criteria (2) Determination of procurement options (3) Collection of utility values for procurement options (4) Collection of selection criteria weightings and a trial run of the model Determination Of Selection Criteria Selection criteria are closely linked with project objectives, which can be tangible such as time and cost, and/or intangible like buildability and relationships. The search for selection criteria was conducted first by reviewing previous studies and then selecting those criteria that were relevant to the local context. A comprehensive list of clients’ requirements can be found in the work of Bennett and Flanagan (p. 26). Hewitt (p. 47) carried out a survey of 21 clients and identified four ‘real needs’: (1) certainty of cost and time; (2) the flexibility to change designs during the construction period; (3) the desire to be actively involved; and (4) innovation from consultants. Findings from the study conducted by Masterman and Gameson (p. 85) highlighted the pragmatic view of clients. In that study, certainty in time, cost and achieving value for money were identified as the primary concerns of construction clients. Other procurement selection models (NEDO, 23; Skitmore and Marsden, 221; Franks, 16) used similar selection criteria. Therefore the selection criteria used in these studies were similar and Table 1 provides the summary. In order to derive a particular set of criteria for use in Hong Kong, a pilot study was conducted with five project management experts who are well recognized practitioners in the field of project management. The experts were presented with the eleven criteria as in Table 1 and were asked to rate the criteria on a Likert scale of 1 to 5 (1 as not important and 5 as very important). The experts were also asked to include other criteria as they considered fit. Only those criteria with mean ratings above or equal to 3 were included in this study. In this regard, eight selection criteria were finally selected. As for the other criteria, they were regarded as less significant. The rating results were reviewed by the project management experts, and they suggested that the primary concerns of construction clients in Hong Kong were related mainly to time, cost and quality, as these would directly affect their profit margins. Although disputes and arbitration were to be avoided, in reality the numbers of disputes that needed to be resolved through arbitration were relatively small. Accountability was an issue for public projects but less so for private developments. Innovative advice from consultants was always welcomed, yet this was usually achieved by way of special contract clauses detailing the procedures and associated time and cost treatments in relation to alternative design proposals. The experts confirmed the rating result, and the eight selection criteria used for this study became: (1) speed: the speed of project completion; (2) certainty: the certainty over the cost for completion of the project; (3) flexibility: the ability and authority of the owner to effect changes; (4) quality level: the quality level required of the completed project; (5) complexity: the suitability of the procurement method in handling complex projects; (6) risk avoidance: the shifting of risk to the contractor; (7) price competition: the degree of price competition associated with the procurement options; and (8) point of responsibility: the clarity of delineation of responsibility. Determination Of Procurement Options The economic boom in the last two decades in Hong Kong produced a great surge in the demand for building and infrastructure developments. The projects undertaken have grown in size, design complexity and construction difficulty. Nonetheless, the development cycle has been shortened as a means of reducing the overall cost of development. The traditional approach of having design fully completed before construction works begin can no longer meet the needs of clients. Various procurement options have therefore been developed to satisfy the needs of the industry. Perry (p. 28) suggests that these options can be categorized generically into ‘design separated from construction’ and ‘design combined with construction’. The classic example of the first category is the ‘lump sum’ contract under which construction work is to commence only after the design has been completed. Practical difficulties render this infeasible in reality. Accelerated traditional procurement has therefore evolved, and construction works commence upon partial completion of the design. The strategies of management contracting and construction management also belong to this category because, as far as design responsibility is concerned, there is no overlap between the consultants and the contractor. With management contracting, the contractor works alongside the design and cost consultants, providing a professional construction management service and subletting the construction works. The management contractor does not undertake either design or direct construction work (Franks, 17). Under a construction management option, the owner engages a construction manager to provide a service for the construction phase, particularly related to the control, management and coordination of the project. All work contractors enter into direct contracts with the client. In this respect, a higher degree of involvement by the owner is expected in managing the interface between design and construction. Design and construct, package deal and turnkey are notable examples in the category ‘design combined with construction’. These options become fashionable when more and more contracting organizations are armed with a design function. In fact, for those with financial capability, their service can be extended to build-operate-transfer. Again with the help of the five project management experts, the following procurement options were identified as those most commonly used in Hong Kong and hence were included in the study. (1) Sequential traditional (full design completed before construction commences) (2) Accelerated traditional (construction commences with design partially completed) (3) Competitive design & build (design & build project secured through competitive bidding. In Hong Kong most D & B projects require the contractor to provide a design in accordance with the clients’ requirements) (4) Turnkey package offered by a contractor. (5) Management contracting (6) Construction management Collection of utility values This stage of the study involves the formulation of a mean utility value for each of the selection criteria against the various procurement options. This process has been adopted in previous similar studies (Skitmoreand Marsden, 223). Formulation of a mean utility factor is aimed at providing a procurement strategy decision-maker with a convenient calculation method when using a procurement strategy decision chart. It is produced so that the decision-maker has no need to produce a custom set of utility factors for each use of the decision chart, hence avoiding biases that may creep in at that point. In this study, a utility factor score on a scale of 10–110 was used to avoid any possible imbalances due to the occurrence of zeros (Fellows and Langford, 37). Data collection was undertaken by way of postal questionnaires sent to 30 developers and 20 project management consultants, randomly selected from the respective directories. Developers and project management consultants were included as they are often involved in procurement option selection. 26 responses (18 from the developers and eight from the project management experts) were obtained. Their mean utility factors were calculated, and are presented in Table 2. The relatively small number of responses warranted an examination of the figures to ensure the acceptability for use in this study. For the criterion of speed, both ‘design & build’ and ‘turnkey’ provide high utility values, which is considered reasonable. It is also within expectation that ‘sequential traditional’ offers the highest utility in terms of certainty of cost. As for the criterion of flexibility, the utility values are generally not high. ‘Management contracting’, in which the involvement of the owner is the highest amongst the six options, offers the highest utility of 74.6. The traditional approach of ‘design then construct’ is often seen as a means of safeguarding quality and price competition. These are reflected in the utilities assigned. ‘Design & build’ and ‘turnkey’ are often used for their ability to shift risk towards the contractors through a single point of responsibility. However, the utility scores for ‘design & build’ and ‘turnkey’ in these two criteria were lower than for the ‘design then build’ options. This may suggest that the owners in Hong Kong are skeptical of the ability of the contractors to assume the design function. As a whole, the utilities assigned fall in line with the perceived order against each criterion. Collection Of Selection Criteria Weightings: AHP Approach Because procurement strategy selection is a decision based on multiple criteria, it is necessary to derive a set of numerical weights representing the relative importance of the criteria with respect to the goal (selection of a procurement strategy). As far as importance weightings for the selection of procurement options are concerned, Love et al. (p. 222) suggest that owners of a similar nature do not necessarily have similar needs. Instead, the needs depend on many factors and often are project specific. As an example, for one project, if the speed of construction (early completion) is the most important aspect, then that client would weigh the selection criterion ‘speed’ higher than the other criteria. This means that deriving a standard set of importance weightings would overlook the project characteristics specific to a particular selection decision. In the proposed model, each decision-maker is to derive his own set of importance weightings for the selection criteria. This enables the project characteristics to be taken into account. Previous studies have used subjective ratings as a means of obtaining importance rankings in the selection of a procurement option (NEDO, 24, Franks, 19; Love et al., 223). Improvement in objectivity is sought in this study by incorporating the analytical hierarchy process (AHP) to assist clients in determining the importance weightings for the selection criteria. The use of the AHP technique enables the decision- maker to structure a complex problem in the form of a simple hierarchy and to evaluate a large number of qualitative and quantitative factors in a systematic manner under multiple criteria. It is a logical way for people to make decisions. Applications of the technique are numerous and these include prioritizing corporate objectives, deciding on products and services to market, allocating funds and resources, prediction and conflict management (Saaty, 9; Saaty and Alexander, 4; Saaty and Vargas, 45). In construction, successful use of the technique has been reported in: prioritizing maintenance schedules (Shen et al., 694); design and build project assessments (Alhazmi and McCaffer, 178) and contractor selection (Fong and Choi, 547). Essentially, the technique employs pairwise comparisons of selection criteria so as to enhance objectivity and downplay too much subjectivity (Saaty, 12). Pairwise comparison forces the decision maker to compare each criterion with all the remaining ones. Table 3 presents the pairwise comparison matrix used in this study. For example, considering the second row, pairwise comparison involves comparing the criterion of speed with that of certainty, then with that of flexibility and so on across the row in a scale of importance. The scale of importance used in this study is also shown in Table 3. As an illustration, if speed is considered to be very strongly important compared with certainty in the selection of a procurement strategy for a project, a ‘7’ is inserted in the juncture cell between speed and certainty. The shaded portion of the comparison matrix need not be completed because these cells should be the reciprocals of the corresponding cells in the non-shaded portion. The ExpertChoice software (ExpertChoice, 17) used in this study handles this task automatically. ExpertChoice is a decision support system (DSS) that enables the decision maker to perform on-screen structuring of a multi-faceted problem in the form of a hierarchy and then allows pairwise judgments to be provided in an interactive and verbal mode (Hokey, 43). The mathematics underlying the use of AHP to generate the relative importance ratings for the selection criteria are founded on linear algebra and graph theory. Details of the mathematical treatment and proof can be found in the publications of Saaty (p. 13). Consistency of the comparison matrix is also important. The term consistency means that when a basic amount of raw data is available then all other data can be deduced logically from it. In other words, it measures how carefully the respondent has completed the matrix. For example, if it is considered that certainty is a factor of 2 more important than flexibility and flexibility is a factor of 3 more important than quality, then to be perfectly consistent, certainty is a factor of 6 (2 x 3) more important than quality. The consistency of the comparison matrix is monitored by an inconsistency ratio (IR) calculated by: IR = Inconsistency index (II)/ Random index (RI) where II = (max – n)/(n-1), with n the number of elements in the matrix (8 in the present study), and RI = the consistency index of a randomly generated reciprocal matrix within a scale of 1 to 9. Saaty (p. 14) gave the RI for a matrix of order 8 as 1.41) and max = the maximum eigenvalue of the comparison matrix. Again, details of the mathematical treatments and proof can be found in Saaty (p. 15). Application of AHP in this study In this proposed method, a decision-maker completes the comparison matrix as in Table 3. The software then calculates the inconsistency ratio (IR). If the IR does not meet the 0.1 threshold, the decision maker then needs to go back to re-examine the ratings assigned and make adjustments where considered necessary. The IR is then checked again. The weighting assignment process is repeated until an IR of 0.1 or lower is achieved. The importance weightings are also calculated. This process ensures sufficient opportunity for the decision maker to exercise his judgments to satisfy the degree of consistency. Through this process, the reliability of the importance weightings can be improved. The results of the importance weightings developed by respondent No. 1 are presented in Table 4. The set of importance weightings for the eight selection criteria represents the degree of relative importance among them, and the sum of the weightings is unity. With the utility values for the six procurement options compiled as in Table 2, the selection model was given a trial run by the research team. Preliminary enquires were made to both private and public owners requesting support, and 15 interviews were conducted successfully. During the interviews, the respondents were first introduced to the concepts underlying the selection model. The respondents, who were regarded as procurement strategy selectors, were asked to consider the last project for which they needed to decide upon a procurement strategy and to assign ratings against the set of selection criteria as in Figure 1. The IR was checked on the spot as described earlier, and this was followed by adjustment of the ratings where necessary. The process stopped when the IR reached 0.1 or lower. With the procurement selection decision chart being organized in a spreadsheet, the utilities were calculated against each criterion in accordance with the importance weightings, and the rankings in accordance with the order of utilities derived were organized. The respondents then compared the proposed option with that actually adopted in their particular project settings. Table 5 presents the procurement strategy decision chart for respondent No. 1. Similarly, 15 rating assessments and hence 15 procurement strategy decision charts were successfully conducted and derived. Table 6 compares the proposed strategies with the actual strategies used. Observations As indicated in Table 6, out of the 15 assessments conducted, there were eight matching and seven non-matching observations. Furthermore, the actual procurement strategies used included ‘sequential traditional’ (six numbers), ‘accelerated traditional’ (seven numbers) and ‘competitive design and build’ (one number), suggesting a seemingly dominant use of the traditional method. To analyze the results and investigate the reasons for the non-matching observations, the importance ratings for each of the selected procurement strategy types were collated. These results are presented in Tables 7–12. In addition, the seven respondents with non-matching results were re-visited in search of an explanation. For the five projects (1,3,6,8 and 15) for which ‘sequential traditional’ was selected as the recommended procurement strategy, there did not appear to be an overriding criterion that had led to its selection. In fact, as shown from the utility table (Table 2), except for speed, flexibility and complexity, the utilities for the other five criteria were close to or above 80. This suggests that the ‘sequential traditional’ strategy is perceived to offer many advantages. It is also important to note that all these five projects were residential developments. In addition, except for project number 6, the other four projects were Government projects. This may be explained by the fact that the designs of public housing have to go through stringent approval procedures to ensure cost efficiency for public accountability purposes. Extensive, repeated uses of an approved design are expected. In this respect, it is quite normal for the design has to be fully developed before construction commences. The ‘accelerated traditional’ category shows a more recognizable pattern. For the two projects (Nos. 4 and 7) in this category, the order and magnitude of the importance of the ratings were fairly consistent. Evidently, from Table 8, speed, price competition, quality and certainty were the four critical selection criteria. The two projects were private residential developments. In view of the high investment cost in the case of Hong Kong, speed of completion can easily be understood as being the prime concern. ‘Accelerated traditional’ is in fact the prevalent procurement strategy being adopted by the private sector. Out of the six projects for which the selection model suggests the use of ‘competitive design & build’ as the procurement strategy, four actually used ‘accelerated traditional’ and one adopted ‘sequential traditional’. This reinforces the observation that the construction community in Hong Kong is entrenched in the traditional approach and reluctant to try new options such as ‘design & build’. The desire to speed up the development process was effected through partial overlapping of the design and construction process. The respondents were asked about their opinions on the divergence between the actual method used and that suggested by the study. For projects Nos. 2, 5 and 13, all private residential developments, the respondents suggested that they were very comfortable with the traditional methods and, unless there were compelling reasons, they were reluctant to try new methods. Furthermore, residential developments in Hong Kong offer little opportunities for input from contractors as the designs are fairly standard and there is little room for prefabrication. For the Government clinic (project No. 12) the respondent claimed that although the Government was advocating the use of ‘design & build’ projects, the clinic was too small in contract value. For project No. 11, which involved institutional quarters, the respondent replied that ‘design & build’ was tried before but the quality of the previous project was not satisfactory. It was that bad previous experience with ‘design & build’ that deterred its use on project No 11. As for the matching project (No. 14), a Government hospital, ‘design & build’ was actually adopted. It was explained by the respondent that the use of ‘design & build’ was the initiative of the Government. Hence it was a policy-driven decision. In actual fact, the number of ‘design & build’ projects commissioned by the Architectural Services Department of the Government of the Hong Kong Special Administration Region constitutes only a relatively small portion of the total number of projects (Table 10). However, in terms of contract value, the proportion is high (Table 11). For example, in 1999, the contract value of the eleven ‘design & build’ projects (approximately 10% of the total number of projects signed) amounts to 46% of the total contract value signed. These figures support the view that ‘design & build’ is used selectively for projects of high contract sum. It has been suggested that the heavier involvement of owners would have a positive effect on project success, especially for complex projects (NWPC, 2). In this regard, management contracting is considered appropriate. This proposition is well supported by the industry experts (Table 2). As projects Nos. 9 and 10 were a hotel and an exhibition centre, respectively, the model seems to have produced sensible suggestions (Table 12). Nevertheless, the actual procurement method used was accelerated traditional, with the respondents admitting that their choice of procurement option was based purely on their usual practice. They found the model useful and provided valuable suggestions to help in the selection process. In summary, the selection model was well received by the respondents. As admitted by the respondents, the selection of a procurement system was largely decided on a judgmental basis. Decisions were made based on the perceptions or previous experience of the decision-maker. The risk-averse attitude of most companies was the inertia that proved dif. cult to overcome. The quantitative approach provided the badly needed supporting information, thereby putting the decision on a firmer footing. The use of the AHP technique requires the respondents to examine all the decision criteria. The checking of the inconsistency ratio also helps to minimize illogical importance weighting assignments. These all help to improve the objectivity of the procurement selection. Conclusion The selection of a procurement strategy for a project is a complex task. The multitude of criteria makes selection difficult. The use of experts to develop the utility factor table was particularly well accepted by the respondents. As for the assessment of relative importance weightings, the analytical hierarchy process improves both the objectivity and the consistency of the weightings. The importance weightings assessment also enables decision-makers to take into account the project characteristics. Testing of 15 sample projects suggests that the method is reliable. The non-matching of the strategies between those proposed by the method and those actually used can be attributed to the ‘habit’ of using a traditional procurement approach and the risk-averse attitude of construction clients in Hong Kong. References Alhazmi, T. and McCaffer, R. (2000) Project procurement system selection model. Journal of Construction Engineering and Management ASCE, 126(3), 176–84. Bennett, J. and Flanagan, R. (1983) For the good of the client. Building, 1 April, 26–7. Edwards, W. (1982) Multi-attribute Evaluation–-Quantitative Applications in the Social Sciences, Sage, Beverly Hills, CA, 1-6. Egan, J (1998) Rethinking Construction, Department of the Environment, Transport and the Regions, HMSO, London, 33-34. Expert Choice (1998), Team Expert Choice: Advanced Group Decision Support Software, ExpertChoice Inc, 14-17. Fellows, R. F. and Langford, D. A. (1980) Decision theory and tendering. Building Technology and Management, October, 36–9. Fellows, R. F., Langford, D., Newcombe, R. and Urry, S. (1983) Construction Management in Practice, Longman, New York, 27-29. Fong, S. W. and Choi, K. Y. (2000) Final contractor selection using the analytical hierarchy process. Construction Management and Economics, 18(5), 547–57. Franks, J. (1990) Building Procurement Systems, Chartered Institute of Building, Ascot. Green, S. D. (1992) A Smart Methodology for Value Management, Occasional Paper No. 53, Chartered Institute of Building, Ascot, 15-19. Hewitt, R. A. (1995) The procurement of buildings. Proposals to improve the performance of the industry, Project Reports submitted to the College of Estate Management for RICS diploma in Project Management, 45-49. Hokey, M. (1992) Selection of software: the analytic hierarchy process. International Journal of Physical Distribution & Logistics Management, 22, 42–52. Latham, M. (1994) Constructing the Team, HMSO, London. Love, P. E. D. Skitmore, M. and Earl, G. (1998) Selecting a suitable procurement method for a building project. Construction Management and Economics, 16(2), 221–33. Masterman, J. W. E. and Gameson, R. N. (1994) Client characteristics and needs in relation to their selection of building procurement systems. In Proceedings of CIB W-92 International Procurement Symposium, East Meets West, Department of Surveying, University of Hong Kong, 4-7 December, pp.79–88. NEDO (1995) Thinking about Building–-A Successful Business Customer’s Guide to Using the Construction Industry, HMSO, London, 23-25. NWPC (1990) No Dispute: Strategies for Improvement in Australian Building and Construction Industry, National Public Works Conference, Canberra, Australia, 1-3. Perry, J. G. (1995) The development of contract strategies for construction projects. Ph.D. thesis, University of Manchester, 27-33. Saaty, T. L. (1990) The Analytical Hierarchy Process, McGraw-Hill, New York, 6-9. Saaty, T. L. (1998) The Analytic Hierarchy Process, Planning, Priority Setting, Resource Allocation. 12-17. Saaty, T. L. and Alexander, J. M. (1999) Conflict Resolution – The Analytical Hierarchy Process, Praeger, New York, 4-6. Saaty, T. L. and Vargas, L. G. (1991) Prediction, Projection and Forecasting, Kluwer, Boston, MA, 45-46 Shen, Q., Lo, K. K. and Wang, Q. (1998) Priority setting in maintenance management: a modified multi-attribute approach using analytical hierarchy process. Construction Management and Economics, 16(6), 693–702. Skitmore, R. M. and Marsden, D. E. (1998) Which procurement system? Towards a universal procurement selection technique. Construction Management and Economics, 6, 71–89. Read More