The ever increasing competition in global markets today has led businesses and companies to find different methods for reducing production and manufacturing costs in order to maintain their competitive edge. The competition has no longer remained company to company but has become supply chain to supply chain. From a buyers perspective a qualified supplier is a key factor to reduce costs. Thus supplier selection and evaluation has gained vital importance in the supply chain management environment. It is extremely essential to develop a supplier selection model which is efficient, effective and considers all the aspects required by the company.
A number of supplier selection methods are available in the current literature. Creating a model based on these methods that addresses the particular requirements of the company is vital. The following paper is in 5 sections. The literature review in the first section is on the various methods for supplier selection and evaluation. The following methods are reviewed.
- Mathematical Programming
- Data Envelopment Analysis
- Analytical Hierarchy Process
- Analytical Neural Network
- Fuzzy Set theory
Along with the review of the methods a discussion on the evolution of supplier selection criteria is also included in the first section. In the second section, two existing supplier selection models in the aerospace sector have been critically reviewed. The description of the Aerospace industry and comparison between the two models is included in the third section. To determine the essential criteria to be included in the model and prioritizing them, the research methodology used was a survey design. The results of the survey are included in the fourth section. The fifth section contains recommendations for building a new model for supplier selection in the aerospace sector.
One of the major topics discussed in most of the production and operations management literature is supplier selection and performance evaluation of suppliers. It is one of the most critical activities of firms due to the increasing significance of the purchasing function (De Boer et al., 2001). The main objective of a supplier selection process is to maximize overall value to the purchaser, reduce the purchase risk and develop a close and long term relationship between the buyer and supplier. Supplier selection is a multi-criterion decision making problem and a number of conflicting factors affect its outcome. The factors taken into consideration are wide ranged and are both quantitative as well as qualitative (Ho et al., 2009). Operational research offers a range of methods and techniques in the form of models which can support the supplier selection decision making. A number of supplier selection methods have been proposed such as data envelopment analysis (DEA), analytic hierarchy process (AHP), mathematical programming, fuzzy set theory and vague set theory, multi attribute rating systems etc. A literature review of international journal articles discussing different multi-criteria supplier selection methods is carried out in this paper. The methods that are prevalently applied in practice, the priority of the evaluating criteria and evolution of selection criteria are also discussed and reviewed. The aim of this paper is to carry out a literature review of the various methods and criteria for supplier selection available in the current literature, in order to produce a set of recommendations for building a new model for the supplier selection process in the aerospace sector. To achieve this, two supplier selection models were critically reviewed, one of which is currently implemented in an aerospace industry and the other one is a theoretical model. A survey on global sourcing and supplier selection process containing 25 questions on various aspects of strategic sourcing was also carried out in order to identify the different characteristics that influence sourcing decisions.
Supplier selection methods:
Supplier selection methods or techniques are the models which are used by decision makers to conduct the supplier selection process. They act as supporting tools for the selection process. The selection of an appropriate method is essential for the overall selection process and can significantly influence the outcome of the selection results (Li et al., 1997). There are number of supplier selection methods available in the literatures.
Mathematical Programming (MP):
MP allows the formulation of the decision problem in the form of a mathematical objective function which needs to be minimised or maximised depending on the objective function by varying the values of the variables. It is an optimization method which selects a number of suppliers in order to maximize either a single criteria or multi criteria objective function subjected to supplier or buyer constraints (DeBoer et al., 2001).
Talluri and Narasimhan (2003) used mathematical programming in the form of a linear programming model to first minimise and then maximise the performance of the suppliers against the best target measures set by the buyers, thus providing a wide-ranging understanding of supplier performance. The authors applied this model considering a set of six suppliers to a Fortune 500 Pharmaceutical company in the process of implementing a JIT system. They regarded price, quality and delivery as the top three criteria for evaluating the suppliers. One of the key features of this max-min approach was that it could identify a set of suppliers with identical characteristics, thus providing the buyer with effective alternates to make their final decision. For the supplier selection problem Ng (2008) developed a weighted linear programming model with an objective function of maximizing supplier score. He implemented the model considering 18 suppliers to a manufacturing firm producing agriculture and construction equipment. He included five criteria namely quality, supply variety, delivery, distance and price. In order to maximise the revenue function Hong et al. (2005) developed a mixed integer linear model to optimize the number of suppliers and order quantity. He applied the model to the supply chain of the agriculture industry in Korea as the customer demand varied seasonally over a period of time. Similarly O’Brien et al. (2001) created a mixed integer non-linear model to optimize the allocation of products to suppliers thus minimizing the annual purchasing costs. Narasimhan et al. (2006) and Wadwa et al. (2007) constructed theoretical multi-objective programming models to optimize supplier selection and order quantity and to minimise lead time, price and number of rejects. Karpak et al. (1999) constructed a goal programming model and applied it to an international manufacturing firm to minimise costs and maximise quality and delivery reliability for selection of suppliers and allocation of products between them. The authors considered cost, quality and delivery reliability as the criteria for supplier evaluation.
On one hand Mathematical programming is advantageous as compared to the other approaches as it takes into account all the constraints during the formulation of the problem. Hence it is much easier to work when a large number of constraints are considered. It can also be used for multiple supplier selection as the current situation can be taken into account in an MP model. On the other hand some of the drawbacks of using an MP model are that it often only considers the more quantitative criteria neglecting the qualitative criteria which are important in supplier selection especially when the goal is to build supplier partnership. Most of the theoretical MP models are complicated to build for the supplier selection problem, due to the large number of variables, but as it can be seen from the above mentioned examples, they can be implemented in an industry as they can be simulated and solved by computers. They are not considered as the most effective method for vendor evaluation as they do not take into account qualitative factors and are incapable of performing a qualitative analysis which is an important aspect of the supplier selection process, thus limiting their use.
Data Envelopment Analysis (DEA):
The concept of DEA is constructed on the basis of calculating the efficiency of the decision alternatives or suppliers. The DEA is a non-parametric method the measures the efficiency without specifying the form of the production function or the weights of different inputs and outputs. The efficiencies are evaluated on the basis of benefits as output and cost as the input criteria (DeBoer et al., 2001). The efficiency of a supplier can be defined as the ratio of the weighted sum of the suppliers outputs to the weighted sum of his inputs, thus the DEA method calculates the most favourable set of weights for each supplier alternative classifying them into efficient and inefficient suppliers. The favourable set of weights that are calculated maximise the supplier efficiency ratings without altering its own rating or making the other suppliers efficiency ratings more than one (DeBoer et al., 2001).
In order to measure the efficiency of alternative suppliers Braglia and Petroni (2000) applied the DEA method by proposing nine evaluating factors to measure the supplier ratings. The authors applied their proposed methodology to the supplier selection process of a middle-sized company manufacturing bottling machinery to test its efficiency. They also calculated the Cross efficiencies in which the weights chosen for a particular supplier can be applied to the weights of the inputs and outputs of the other suppliers as well as Maverick index which is the percentage relative difference between cross efficiency and simple efficiency in order to avoid the selection of false positive supplier. Talluri and Barker (2002) and Talluri and Sarkis (2002) applied DEA to evaluate suppliers, manufacturers and distributors as a three phase approach for a logistics distribution network. They also employed the DEA to measure the performance of the suppliers using six evaluating factors having two inputs and four outputs. Ross et al. (2006) evaluated the supplier performance with respect to the performance attributes of both buyer and supplier by using DEA. The author carried out three sensitivity analysis; the first one computed supplier efficiency scores without taking into account the evaluation team’s and the buyers weights. The second analysis considered the evaluation taking into account the team’s preferences and the third analysis considered the buyers preference.
Liu et al. (2000) constructed a DEA model to evaluate the overall performance of a supplier considering three inputs namely price index, delivery performance and distance factor and two outputs which were supply variety and quality. The authors applied the DEA model to a firm manufacturing agriculture and construction equipment containing a multi modal assembly line. The model could select suppliers with a high supply variety, thus reducing the number of suppliers. Seydel et al. (2006) developed a DEA model to evaluate technology suppliers considering three factors. He included amount of know-how transfer as a qualitative factor in the model. The author developed a five point scale to rank the suppliers in term of the qualitative factor.
The DEA method provides a means to evaluate and select suppliers on the basis of their performance over a period of time. It compares supplier performance in a multi criterion setting thus allowing the purchasing firm to evaluate each supplier’s performance relative to the performance of the best supplier in the market by calculating the efficiency measures. Observed supplier performance data is used in a DEA method, thus the purchasing firm does not have to calculate its own utility functions as is required in the other techniques. Some of the limitations of the DEA approach are that its focus is not on selection an optimal supplier as the other mathematical programming models; hence it cannot be used if the purchasing firm requires the selection of an optimal supplier. The DEA model also makes some assumptions like any other supplier selection model thus limiting its use. (Garfamy et al., 2006)
Evolution of the Analytic Hierarchy Process (AHP):
Linear Weighting Model:
Weights are assigned to the criteria with the largest weight corresponding to the highest priority, in a linear weighted model. The ratings of the criteria are then multiplied with their respective weights and the sum of weights is assigned to each supplier, thus the supplier with the highest overall rating can be selected. There are a few imprecision’s in the rating mechanism such as difficulty to determine the score of a supplier on a criterion or importance of some criterion with a high degree of precision. To overcome these imprecision’s the use of analytic hierarchy process (AHP) was proposed (DeBoer et al., 2001).
Analytic Hierarchy Process:
The AHP is a decision making method first introduced by Saaty (1980) which prioritizes alternatives or suppliers when considering multiple criteria, thus allowing the decision maker to restructure complex problems in the form of a set of integrated levels or a hierarchy. It is one of the most commonly applied methods in practice as it incorporates qualitative as well as quantitative criteria and is relatively simple to understand. Various adaptations of AHP have been developed since its introduction.
Muralidharan et al. (2002) developed an AHP model consisting of five stages to rate and select suppliers considering nine criteria. Some of the major criteria that the author considered were quality, delivery, price and technical ability. The model was then applied to the supplier selection process to evaluate six suppliers of a leading organization manufacturing bicycles. Liu and Hai (2005) created an AHP model and used Noguchi’s voting and ranking system thus allowing each manager to determine the order of criteria instead of weights for the selection and evaluation of suppliers. They used a six step process for supplier evaluation and considered eight criteria in their analysis, some of them being quality, responsiveness, delivery, technical capabilities etc. The authors applied this model for selecting one of ten suppliers for the Umbrella Scheme of Malaysia’s furniture industry.
Chan and Chan (2004) constructed an AHP model considering six criteria namely cost, delivery, flexibility, innovation, quality and service with twenty sub-factors among them. They applied the model to the supplier selection process of a leading company that manufactures and supplies semiconductor assembly equipment assuming that the supplier had to be chosen for a critical product. The relative priority ratings were calculated based on customer or buyer requirements. Hou et al (2007) developed a decision support system based on AHP in a mass customization environment considering internal and external factors to meet market requirements. The author applied the model to the selection process of a subsidiary company of a local Chinese printer manufacturer. Chan (2003) created an AHP based interactive selection model which determined the relative importance of evaluating criteria without being subjected to human judgment. The AHP model can also be integrated with other supplier selection models in order to achieve optimized selection results.
Ramanathan (2007) suggested that the qualitative and quantitative information gained from the total cost of ownership model and AHP model can be utilized to evaluate the performance of a supplier using the DEA method. The author considered the costs from total cost of ownership as inputs and the weights gained from the AHP method as outputs. Sevkle et al. (2007) applied the AHP-DEA integrated method to solve the supplier selection problem of a major Turkish TV manufacturer BEKO, in which he used AHP to derive local weights from a given comparison matrix and summed up the local weights to get the overall weights. In order to calculate the efficiency scores of all the suppliers DEA was used on the decision making units. Percin (2006) applied integrated AHP-GP method, where AHP was used to measure the priority weightings of alternate suppliers considering twenty evaluating factors. The author used the weightings obtained by AHP Goal programming method as the coefficients for five objective functions. The integrated model was used to optimize the order quantity from the most appropriate supplier considering the capacities of the suppliers.
Mendoza et al. (2008) offered an integrated AHP- GP model in order to reduce a large number of potential suppliers to a manageable figure. He ranked the alternatives considering five evaluating criteria to optimize the order quantity. Xia and Wu (2007) applied the AHP model to calculate the performance scores of potential suppliers. The authors then applied the scores as coefficients of one of the four objective functions in a multi-objective mixed integer programming model. The model was developed in order to determine the optimal number of suppliers and to select the best set of suppliers.
Some of the advantages of AHP method are as follows (Chan et al., 2003)
- The system can be represented in a hierarchical manner to explain the changes in priority and its effect at upper and lower levels.
- The desired performance of the supplier is characterized by hierarchical selection criteria viz. the management of the suppliers is better if the suppliers performance is evident to the buyer.
- It utilizes multiple paired comparisons of criteria to rank order alternatives and it is the most exceptional Multi-criterion decision making approach.
- Efficiently progresses through modular construction and final assembly of modules than those assembled as a whole, this is known as hierarchical assembly of natural systems.
- Identifies the key elements assisting in making more accurate business decisions and is a structured method which obtains information from target respondents (decision makers or experts).
- It provides information regarding the structure and function of a system in the lower levels of the hierarchy and gives the outline of the criteria and their purposes in the upper levels. Limitations on the elements in a level are best denoted in the next higher level to ensure they are satisfied.
- It has stability and flexibility, stability as small changes have small effects and flexible in the sense that the performance is not hampered if there are any additions to a well structured hierarchy
Disadvantages: (Chan et al., 2003)
- Most of the supplier selection problems do not have a single hierarchy.
- Utilization of this statistical method is complicated for most of the users and this makes the process unmanageable.
- It is not cost effective to procure the essential information i.e. due to lack of information /willingness to compare two alternatives with respect to some criterion the supposition of comparability is invalid.
- To reach an agreement with the team members by reviewing the models is time consuming.
- The presumption that the relative importance of criteria affects the supplier’s performance is definite which cannot effectively take into account the risk and uncertainty in assessment of supplier’s potential performance.
Analytic network process (ANP):
Sarkis and Talluri (2000) suggested the use of analytic network process, which was a more sophisticated version of the AHP method. The authors believed that the supplier evaluating factors could influence each other and this interdependency needed to be considered in the process. They applied the ANP process to evaluate and select suppliers in a company manufacturing custom-designed high technology metal-based products, considering organizational factors and strategic performance matrix. The model included seven evaluating criteria namely cost, quality, flexibility, delivery time etc. also considering their interdependencies. Bayazit (2006) implemented an ANP model considering ten evaluating criteria. Some of the important criteria included were on time delivery, quality, flexibility and delivery lead time. He classified the criteria into supplier performance and capabilities clusters and the interdependencies among them were formulated by considering each cluster as a controlling factor for a pair wise comparison matrix.
Demirtas and Ustun (2008) developed an integrated model in which they used ANP to evaluate the performance of potential suppliers considering 14 criteria. The weights were then considered in one of the three objective functions of a multi-objective mixed integer programming model. Similarly the authors integrated the ANP and the GP methods of supplier selection and evaluation in 2009. The only difference to the previous model was that there were four goals in the GP model. Gencer et al. (2007) developed ANP model considering various evaluating criteria. He classified them into three clusters to take into account their interrelationships to evaluate and select the most appropriate supplier.
Some major advantages of ANP process over AHP are that ANP provides with additional insight as most of the real world supplier selection problems have interdependencies among the evaluating criteria. It also incorporates both qualitative as well as quantitative factors which are important in supplier selection. The ANP method can deal with various uncertainties and complexities as it makes use of ratio scales to incorporate a variety of interactions. In spite of the advantages, the ANP method does have a few limitations as it is a very complex method and requires additional effort and time as compared to AHP.
Fuzzy Set Theory:
The fuzzy set theory is used to model uncertainty and imprecision in the supplier selection situation. Fuzzy set systems make use of linguistic rules which are very well suited to describe the behavior of practical problems. In most of the real world applications, fuzzy rules are created by the decision makers with a few input variables. When the number of input variables increases, the possible number of fuzzy rules for a particular system increases exponentially. It is rather difficult for the decision maker to generate a complete set of rules to assess the supplier selection system (Chan et al., 2006).
Chan et al. (2006) presented a hierarchy model based on the fuzzy set theory which could deal with both quantitative and qualitative criteria. The author used linguistic values to assess the ratings and the weights for the evaluation factors. The ratings were arranged in triangular fuzzy numbers. They created a hierarchical structure of the decision problem and applied the model to a high technology manufacturing company to select a suitable supplier to supply material for key components of a new product. Sarkar and Mohapatra (2006) used a fuzzy set method to eliminate the imprecision in a number of subjective characteristics of suppliers. The authors evaluated and selected the suppliers on the basis of performance and capabilities as the two major measures for evaluation. They considered a hypothetical case to exemplify their model by considering a pool of ten suppliers and the goal being to reduce that number and select the best two suppliers.
Kahraman et al. (2003) applied the integrated fuzzy AHP approach to select the most appropriate supplier for the biggest white goods manufacture in Europe to supply the plastic part scroll housing for their new model of aspirators. In this model the decision makers could specify their preferences in terms of linguistic variables regarding the priority of each evaluating criteria. Chan and Kumar (2007) also applied a fuzzy AHP methodology for selection of suppliers. The authors used triangular fuzzy numbers and fuzzy synthetic extent analysis methods to choose the final priorities of different criteria. The authors applied the model to the supplier selection process of a manufacturing company to select the best global supplier for one of their critical parts used in the assembling process. The criteria considered in the model for evaluation were overall cost, quality of product, service performance, supplier profile and risk factors.
Amid et al. (2006) formulated an integrated fuzzy multi-objective linear programming model which took into account the vagueness and imprecision of the input data in order to optimize the order quantity. The author developed an algorithm to solve the model which incorporated three objective functions with different weights. They considered a hypothetical case to select three suppliers for supplying a new product to a market. The purchasing criteria considered for the model were net price, quality, service and capacity. The author also formulated a fuzzy multi-objective mixed integer programming model which was similar to the earlier model but it also took into account the quantity discount. The price discount was directly proportional to the quantities ordered (Amid et al, 2006).
One of the primary advantages of using fuzzy set theory for supplier selection is that it makes use of linguistic variables, which are highly beneficial when the performance values cannot be expresses in terms of means of the numerical values. Thus, taking into consideration the uncertainty and imprecision of the quantitative data gathered by the purchasing company or provided by the supplier. It is beneficial and easier to use linguistic variables instead of numerical values while assessing potential suppliers with respect to criteria and weights. A modified fuzzy set theory is capable of handling both qualitative as well as quantitative data ratings and is flexible in use, which is an added advantage (Chan et al, 2006). Some of the disadvantages of fuzzy set theory are that the analysis is based on the theory and not exploratory data; hence validation of the data may be required. It is a subjective methodology, thus justification for each step is necessary. As the number of variables increase the complexity increases, thus requiring a number of procedures in the sub-systems of the methodology.
A number of other methodologies exist for the supplier selection problem such as artificial intelligence and expert systems which includes case based reasoning (Choy et al, 2005; 2002; Humphreys et al, 2003) and Bayesian belief networks (Kreng et al, 2003). Multi-criteria decision methods which include outranking methods (DeBoer et al, 1998; Dulmin et al, 2003), judgmental modeling (DaSilva et al., 2002; Naude and Lockett, 1993), interpretive structural modeling (Mandal and Deshmukh, 1994) and categorical methods (Houshyar and Lyth, 1992). Multivariate statistical analysis that incorporates structural equation modeling (Lin et al., 2005; Tracey and Tan, 2001), Factor analysis (Krause et al., 2001; Tracey and Tan, 2001) and confidence interval approach (Muralidharan et al., 2001). Group decision methods (Han and Ahn, 2005; Mandal and Deshmukh, 1994) and multiple integrated methods also exist for supplier selection.
All the methods that are utilized for selections of suppliers have their own advantages and disadvantages. No method can be said to be the perfect method which covers all aspects of the entire selection process. Modifications and improvements can be made to every method in according to the requirements of the decision makers. The selection process can be improved by integrating different techniques in order to negate the limitations of the techniques taken individually. Considering this procedure, the fuzzy integrated AHP model and the DEA integrated ANN model are comparatively the best combination of methods that can be implemented for supplier selection.
Supplier Selection Criteria:
Evolution of supplier selection criteria:
A number of criteria need to be considered for the supplier selection decision making process which makes the selection of suppliers a complicated practice. Since the early 1960’s, practitioners and academics have been focusing on the analysis of supplier selection criteria and measurement of supplier performance. Dickson et al, (1966) suggested “From the purchasing literature is fairly easy to abstract a list of at least 50 distinct factors that are presented by various authors as being meaningful to consider in a vendor selection decision”Â?. In his work he carried out a survey to identify the most important criteria required for the selection of suppliers. The author came up with 23 criteria and their relative importance for vendor selection. The following table summarizes the 23 criteria and their level of importance.
Weber et al. (1991) conducted a similar study on the bases of the 23 criteria identified by Dickson (1966). The authors reviewed and classified 74 related articles appearing between 1966 and 1990. Their study provided a clear indication of the issues concerning selection of suppliers. Both the studies indicated net price, quality, delivery and production facility and capacity as the top 4 criteria for supplier evaluation. These two studies were the primary studies done on supplier selection criteria and were the bases of a number of papers in the forthcoming years.
A number of changes at a profound level have taken place in the business environment, including purchasing and procurement since Weber et al.’s work in 1991. The basic definitions of Dickson’s 23 criteria have undergone change and expansion and new criteria have emerged due to a substantial growth in business and supply chain needs. Dickson (1966) defined net price as price offered by each vendor including discounts and freight charges. In the development of the net price criteria, the term net price had been replacement by the term cost which includes a number of costs such as fixed cost, inventory costs, ordering costs, supplier costs and costs associated with quality, after-sales and technology (Current and Weber, 1994). The term total cost of ownership has also become important in recent times which include not only the purchasing price but also purchasing related costs (Bhutta et al, 2002).
The delivery criterion was defined by Dickson (1966) as the ability of each vendor to meet specified delivery schedules. The delivery criterion has now been developed to incorporate lead time, cycle time, shipment quantity and quality, delivery capacity etc (Karpak et al, 1999). According to Dickson quality was defined as the ability of each vendor to meet quality specifications consistently. The quality criterion has now been extended to include inspections and certain specifications such as the ISO9001 system (Lee et al, 2003)
In addition to the evolution and development of the basic criteria a number of new criteria have emerged in literature from various authors. Some of the new criteria are flexibility, which includes process and production flexibility, response to change, responsiveness to customer needs (Ghodsypour et al, 2001), flexibility to change the order and order quantity and ability to respond to fluctuating demand (Verma et al, 1998). A product design and development criterion consists of commitment to continuous improvement, product development and improvement, design capabilities and continuous improvement in product and process (Chan et al, 2003). Supplier relationship is another criterion that has gained importance in recent years due to integration of various sections of supply chain. Supplier relationship has two aspects, strategic and tactical. The criterion can be sub divided into 4 sections namely strategic long term relationship, tactical long term relationship, strategic short term relationship and tactical short term relationship. Due to the growth in the businesses, buying firms prefer to integrate the suppliers in their supply chain, thus forming a strategic long term alliance with their supplier