Determining Strategy Based Supplier Pre-Qualification Criteria With Fuzzy Relational Maps

Supplier Selection is one of the most studied areas in management and decision sciences. However, it is still a highly problematic subject since decision-makers take an educated guess after a certain stage in real life practices. In order to overcome the problems, decision makers should focus their attention on the first stage of the selection process, criteria determination, as the quality of the selection phase heavily depends on the first stage. Additionally, strategic fitness in supplier selection is also not much considered in the literature and real life practices. However, looking for conformity of supplying organizations with corporate strategies of buying organization is crucial for the success and leadership of the buying organization. Therefore, this paper intends to determine the most influential corporate strategy based supplier pre-qualification criteria. Data acquisition phase of the Delphi technique was used for determining criteria and fuzzy relational maps was used for relating criteria with corporate strategies. All data were collected from a global Tier-1 manufacturing company in the automotive industry. The results show that the most important strategy based criteria were mainly about organizational and managerial characteristics of the company. Cost and price, which are considered very important in the literature and in real life practices, were determined as moderately important in strategic context. Companies need to focus their attention on criteria such as technical qualification of employees, continuous improvement systems, and communication abilities when pre-qualifying suppliers


Introduction
Dickson was undoubtedly the pioneer in the subject of supplier selection, introducing the first set of criteria for supplier selection (SS) in 1966.However, one of the earliest records about procurement and supplier selection belongs to Lewis in 1933, in which he states "There is none more important than the selection of a proper source.Indeed, it is in some respects the most important single factor in purchasing" (Vokurka, Choobineh, & Vadi, 1996).Therefore, one can see that the importance of the selection of a proper source and purchasing was recognized in the early 1900s and its importance today is still rising.As cited in Vonderembse and Tracey (1999), in most manufacturing organizations, the role of purchasing and materials management has gained viability and additional responsibility.According to the authors, in some organizations, such as Honda of America and Daimler-Chrysler, it has gained recognition from top management as a key process (Vonderembse & Tracey, 1999).
Supplier selection is defined as a major element (Raut, Bhasin, & Kamble, 2010), essential activity (Chang, Chang, & Wu, 2011) and most capital decision (Benyoucef, Ding, & Xie, 2003) in the global economy because of following factors: Globalization (De Boer, Labro, & Morlacchi, 2001), increase in costs of outsourced materials (Aissaoui, Haouari, & Hassini, 2007), expanded supply chain networks (Benyoucef et al., 2003), rapid and continuous change (Azadi & Saen, 2012) and eager to gain competitive advantages (Moser & Blome, 2008).All of these factors make supplier selection very complicated because they inherit some paradoxes that require use of conflicting quantitative and qualitative criteria in the selection step of the process (Aissaoui et al., 2007).Moreover, due to vague nature of qualitative criteria, ratings of decision makers about same supplier(s) may differ.This brings more complexity into the process.
In the literature and also real life practices, there is quantum of work has been done in supplier selection.However, many researchers and practitioners still miss a key point.They mainly focus on the selection step which is the last step of the whole process (R. Jain, Singh, Yadav, & Mishra, 2014).However, it is crucial to put an effort on previous steps, criteria determination and pre-qualification because they affect quality of the final decision (De Boer et al., 2001).Another overlooked point in the selection is the strategic interfirm-fit between the subject company and its suppliers.Toulan et al. (2006) state that greater fitness between organizations and their strategies will foster better the performance of relationships between organizations (Toulan, Birkinshaw, & Arnold, 2006).Moreover, strategic congruence between organizations is a key to stay in the competition (Nilsson & Rapp, 2005) so to have sustainable leadership (SL).
Due to all of aforementioned factors, this research aims to introduce a hybrid approach incorporating fuzzy relational maps (FRM), which is a more developed style of fuzzy cognitive maps (FCMs), with Delphi technique to create a set of strategy based criteria that can be used in pre-qualifying suppliers in order to determine strategically misfit companies.Application of the approach was examined in a case of a global Tier 1 manufacturing company operating in automotive industry.
Alphanumeric Journal Volume 4, Issue 2, 2016 Supplier pre-qualification term is used interchangeably with the terms supplier initial screening, supplier shortlisting, supplier pre-evaluation and supplier pre-selection in the literature.
The remainder of this paper is organized as follows: Section 2 discusses the related literature about today's supplier selection environment, supplier pre-qualification, pre-qualification criteria and strategy based supplier selection/pre-qualification criteria.Section 3 describes a review of Delphi technique and fuzzy relational maps.Section 4 introduces the approach and its application.Finally, section 5 summarizes results and discussion.
In order to lower costs, organizations tend to do some purchases from foreign suppliers from low cost countries (Cheraghi, Dadashzadeh, & Subramanian, 2011;Pal et al., 2013;D. D. Wu, Zhang, Wu, & Olson, 2010).However, outsourcing from low cost countries poses communication, delivery and quality risks.
 The cost of outsourced materials: It is enormous in today's business due to high outsourcing rates (Aissaoui et al., 2007;Benyoucef et al., 2003;Micheli et al., 2008).For example, several decades ago, the cost of outsourced materials and services comprises up to 80% of the total, especially for high-tech products (Burton, 1988), and for today, it was determined specifically 70% of total operation cost of an automobile (Yu & Wong, 2014).
 Expanded supply chain networks: It is a result of high outsourcing rates (Benyoucef et al., 2003).Expansion is something desired under today's market trends.On the other hand, it may be too dangerous to have numerous suppliers because as the reliance to suppliers grows, performance of the subject company greatly depends on actions of suppliers (Vonderembse & Tracey, 1999).The dependency makes the chain more susceptible to disturbances (Nepal & Yadav, 2015) that may increase total cost and tarnish reputation and position of the company as well (P.-S.Chen & Wu, 2013).
 Competitive advantage: SCM activities have been forced to change its traditional role from a provider of the right products/services at the right time and lowest costs to a generator of competitive advantages (Moser & Blome, 2008;Vonderembse & Tracey, 1999).Activities of suppliers provide value for the subject organization and its customers as well (Siguaw & Simpson, 2004).Therefore, organizations have to work with their supply chain (SC) partners to stay in the competition and improve total performance of the chain (Aissaoui et al., 2007).
All of these factors make supplier selection very hard.
A brief definition of supplier selection is the process by which candidate organizations are reviewed, evaluated and chosen to become a part of supply chain of the company (Aghai, Mollaverdi, & Sabbagh, 2014).The outcomes of the process provides the most suitable suppliers which are able to provide the subject company with the right products/services at the right price, in the right quantities and at the right time (You, You, Liu, & Zhen, 2015).An extensive literature review and detailed information about supplier selection and its methods can be found in (Chai, Liu, & Ngai, 2013;De Boer et al., 2001;Degraeve, Labro, & Roodhooft, 2000;Ho, Xu, & Dey, 2010;Karsak & Dursun, 2016;Pal et al., 2013;Surajit Bag, 2011;Weber, Current, & Benton, 1991).Even though, voluminous work has been published about this subject, there is not much specific paper about the strategic criteria determination and pre-qualification (Aissaoui et al., 2007;Arikan, 2013;De Boer et al., 2001;Ho et al., 2010;Shaw, Shankar, Yadav, & Thakur, 2012).

Supplier pre-qualification
As cited in Lee (2009) gathering valuable information about resources, capabilities, motives, management and reliability of candidate partners requires a very long time.
Otherwise, rushing into a partnership with inadequate preparation often leads to the failure of relationships (Lee, 2009), and the selection of an inadequate supplier can produce disastrous results for the subject company (Wilson, 1994).Therefore, there is a need for a much systematic approach (Aissaoui et al., 2007;Y. H. Chen & Chao, 2012).
An efficient and flexible supply chain enables firms to select right suppliers at the right time for the right purpose, to provide product/services more effectively and efficiently.It is not only significantly reducing purchasing cost but also greatly improving customer satisfaction and corporate competitiveness (Florez-Lopez, 2007; R. Jain et al., 2014;Kotula, Ho, Kumar Dey, & Lee, 2015;Lin, 2012;Paul, 2015).If the selection is done properly, a better quality and long lasting supplier relationship is attainable (Igoulalene, Benyoucef, & Tiwari, 2015;R. Jain et al., 2014).These relationships can be obtained only if the number of suppliers is reduced/shortlisted to a small number of qualified companies (Aissaoui et al., 2007;Swift, 1995;Wilson, 1994;Yu & Wong, 2014).However, it is a very hard task to achieve.Therefore, a supplier selection model including supplier pre-qualification may be an advantageous approach especially if there are a large number of suppliers (Spekman, 1988;Yu & Wong, 2014).
According to (De Boer, L., Labro, E. e Morlacchi, 2001), supplier pre-qualification is one of the initial steps of classical supplier selection process.It is defined as a process of reducing/shortlisting the number of candidate suppliers to a small set of acceptable suppliers.It is defined as a classifying process rather than a ranking process (De Boer et al., 2001).In other words, while supplier selection is about candidacy of the most possible suppliers, supplier pre-qualification is about nomination candidacy of possible suppliers.It is a step that reduces the number of possible suppliers to a small group of them.However, to the best of our knowledge, pre-qualification is usually omitted both in literature and real life practices (C.G. Şen et al., 2010).To the best of our knowledge, there are just a few research papers about supplier pre-qualification (Cao et al., 2014;Luo, Wu, Rosenberg, & Barnes, 2009;Sarkar & Mohapatra, 2006;C. G. Şen et al., 2010;Yu & Wong, 2014).

Pre-qualification criteria
The decision process is a time consuming task (Lee, 2009) because all of aforementioned factors bring some paradoxes with.For example, it is hard to decide which one of the following alternatives shall be rational for a company; working with one, some or many supplier, buying from a low cost or high quality supplier (Aissaoui et al., 2007)?Definitely there is no answer that fits all procurement cases.Because of this complicated nature of the selection, types of criteria used in the decision process also get complicated.It is important to use both quantitative and also qualitative criteria to make a proper decision.As a result, some criteria may conflict each other (Aissaoui et al., 2007;Benyoucef et al., 2003;R. Jain et al., 2014).Therefore, working with both quantitative and qualitative criteria at the same time makes decision making process a bit harder.Moreover, experiences and feelings of decision makers are involved in the decision process with the inclusion of qualitative criteria.Consequently, it becomes necessary to make a trade-off between conflicting tangible and intangible criteria (R. Jain et al., 2014).Criteria used in supplier pre-qualification articles mentioned in Section 1.2 are listed in Table 1.To the best of our knowledge there is only one article about determination of supplier pre-qualification criteria belongs to (R. Jain et al., 2014).The authors propose to use data mining technique integrated with fuzzy association rules approach.They aim to investigate the influence of suppliers' initial strength on their final work.During the investigation process, they also propose a set of supplier pre-qualification criteria which were actually collected from the literature by the authors.Based on the finding of this paper, organizational strength, performance capabilities and miscellaneous are the most important criteria that affect the final selection.Volume 4, Issue 2, 2016 On the other hand, some articles about supplier pre-qualification have subsections on the criteria determination.Please see Table 2 for the techniques used for it.In addition to these articles, in a literature review article of (De Boer et al., 2001), interpretive structural modelling and expert systems are defined as proper techniques for criteria determination.
Operating like an autonomous individual company is no longer accepted.Instead, establishing a sounder strategic alliance with suppliers against competitors is more desirable (Shen & Yu, 2012;You et al., 2015).Therefore, building and maintaining close and long term relationships with a few albeit reliable and high-quality partners (Aissaoui et al., 2007;Y. H. Chen & Chao, 2012;W. Y. Wu, Sukoco, Li, & Chen, 2009;You et al., 2015) is a must to stay in the competition and obtain leadership.
To explain, supply chain activities require coordination, collaboration and cooperation between partners of the chain to maximize the efficiency (Igoulalene et al., 2015;Swift, 1995).This harmony in the chain helps to build and strengthen long standing mutual interactions which are vital for the total value of the chain.It provides synchronization of activities in group members, binding hidden force that unifies actions of the members and mutually decided common goals for the members (Ertay, T., Kahveci, A. & Tabanlı, 2011).However it is not easy to create common goals, synchronization and unification if there is no congruency between the subject company and its vendors.
According to Richards & Jones (2007), there are three types of congruency among organizations, strategic, operational and personal (Richards & Jones, 2007).As cited in Toulan et al. (2006), these types of fitness are actually based on Miles & Snow (1978) seminal work about organizational strategy.According to the authors, greater fitness between organizations and their strategies will result in better performance of the relationship (Toulan et al., 2006).Strategic congruence is a necessary precondition for being competitive (Nilsson & Rapp, 2005).Otherwise conflicting organizational objectives/strategies/values may cause inefficiencies between partners (As cited in Siguaw & Simpson, 2004).In other words, procurement decisions should support the common goals of the partnership and organizational strategies of the subject company (Burton, 1988).Having the best supplier may not guarantee "the best relationship" if the supplier does not consider the business direction of the Alphanumeric Journal Volume 4, Issue 2, 2016 subject company.As a result, the evaluation criteria must be in alignment with organizational strategies of the subject company (Shen & Yu, 2012).However, there is not much research paper about strategy based supplier selection or prequalification criteria.
Sarkis Talluri (2002), used a set of strategic performance metrics (and also organizational factors) in order to evaluate alternative suppliers strategically (Sarkis & Talluri, 2002).In the study of (S.Şen, Başlıgil, Şen, & H., 2008), supplier selection criteria and their importance weights were determined based on buyer-supplier integration levels.A five level integration model of Ghodsypour & O'Brien (1988) was used for determining importance weights of criteria.In the study of (Chou & Chang, 2008), a supplier selection model that had a criteria, "strategic fit" was proposed.It represented "the fit between firm strategy and supplier strategy".However, the importance weight of this criteria was judged by the decision committee in the buying company.The final selection was made in the light of expected collaboration type with the supplier.Finally, (Shen & Yu, 2012) proposed a set of business process improvement oriented strategic criteria, which were named as process capability indices modified from Garfamy (2005), for supplier selection.Please see Table 3 for the criteria used in these articles.
References (Sarkis & Talluri, 2002) (S.The Delphi technique is designed to overcome interpersonal behavioral problems in group decision making (Azadeh, Keramati, & Jafary Songhori, 2009).A typical consensus process consists of repeated interrogations / Delphi rounds , usually three or four, through questionnaires (Azadeh et al., 2009;C.-M. Wu et al., 2013).However there might be some exceptions.For example, as Fischer stated, there are several variations on the method.In order to reduce the number of rounds, it is possible to prepare the first questionnaire with the director(s) before the first round (Fischer, 1978).Also as group members do not have any face to face meeting during the process, disadvantages of group decision-making, such as groupthink, group polarization and dominant character, are removed from the process (Dalkey & Helmer, 1963).Each interrogation in the process is accompanied by group replies belong to proceeding round.By doing so experts are encouraged to reexamine and (if necessary) change their previous reply in the light of proceeding replies of other group members (W.Y. Wu et al., 2009).
Nodes of the maps represent concepts, indicating the main features, nature or attributes of the system (Xiao, Chen, & Li, 2012).Their edge connections represent causality between the nodes (concepts).This structure also allows users creating new maps by combining several maps (Glykas, 2010).Therefore, they are a practical tool for working with a large quantity of data.Moreover, fuzzification enables working with qualitative and quantitative data between which dependencies do not fit in classical logic (Kandasamy & Smarandache, 2003;Zadeh, 2008).Therefore, it is important to note that matrices of FCMs are always square matrices of which diagonal entries are zero.
Fuzzy relational maps (FRM) are more developed style of FCMs.As (Kandasamy & Smarandache, 2003) stated "In FCMs we promote the correlations between causal associations among concurrently active units.But in FRMs we divide the very causal associations into two disjoint units".In other words while FCMs cannot provide the effect of one group on another group, FRMs can give that effect.Therefore, contrary to the structure of FCMs, there is a need for a disjoint sets in the sense of concepts.These sets are called domain and range space.Number of nodes in the range space do not need to be equal to the number of elements in the domain space.].
FRMs also enable to inter-relate two separate systems (Kandasamy & Smarandache, 2003).To explain, suppose there are three spaces, assigning numerical weights to connections of FCMs (Glykas, 2010) which is applicable to FRMs as well.

Framework for Defining Strategy Based Criteria and Their Influence On Competitiveness
A systematic procedure consists of Delphi technique and fuzzy relational maps is proposed for this research.The research procedure consists from two main parts, criteria determination and influence estimation.The Delphi technique was used for extracting knowledge about supplier pre-qualification criteria, and fuzzy relational maps for estimating its influence on corporate strategies and the goal of the company which was sustainable leadership.The proposed framework was applied in a Tier-1 company operating in automotive industry in 19 countries over the globe.It develops and manufactures car interiors and also driver and passenger seating systems for off-road vehicles, trucks, buses, and trains.It has an extensive supply network in which 600 suppliers.
 Step 1. Forming a decision unit: A list of experts is created initially.At the beginning of the research, there were one procurement director and five procurement executives.However, during the research, two executive left the company; and the research was done with one procurement director and three procurement executives.


Step 2. Compiling a list of criteria and preparing the first questionnaire: After that, a questionnaire explaining nature of the study and asking each person to generate a list of important criteria is sent to the experts.However, in this research as stated in section 3.1., in order to reduce the number of rounds (Fischer, 1978) and to inhibit experts' burnout (Siguaw & Simpson, 2004) the first questionnaire was created with the procurement director of the company.After making a literature review regarding supplier selection and pre-qualification criteria, a meeting was hold with the procurement director, and all possible criteria were determined.However, some criteria were replicated to determine the evaluators' attention level since the number of criteria was very high.Eventually, the list was compiled, the questionnaire was created in MS Excel, and e-mailed to the experts.They were questioned regarding the importance of criteria to purchasing context.All items in the questionnaire were evaluated by using a five point Likert scale where "1" denoted not important and "5" denoted absolutely important.Also they were asked to propose new criteria for the next round if they thought any significant criteria were overlooked.


Step 3. Evaluating the results of the first round: From the first round responses median and interquartile range is calculated.After that results of the first round is returned to the evaluators as a feedback, asking them to consider their responses if they are out of the range or to state reasons for remaining out of the range.In this research, all necessary actions were completed.Additionally, all repeated questions were discarded.Redesigned questionnaire were emailed back to the decision makers.


Step 4. Evaluating the results of the second round: The process of the first round is repeated and respondents are asked for a final revision of their responses.In this research, median and interquartile ranges were calculated again, and final revisions were requested from the experts.


Step 5. Preparing the final report: The outcome of the rounds is a list of possible important criteria with respect to supplier selection context.In this research, the list was finalized and sent to the procurement director for his information.
Additionally, on the request of the procurement director, all criteria grouped under 10 headings in order to ease the evaluations that will be made in the next step.
 Step 6. Collecting corporate strategies of the subject company: In this step, existing corporate strategies of the organization was outlined.There were seven of them, using cutting edge technology, satisfying customer needs, making environment friendly production, retaining existing quality certificates and obtaining new ones, keeping supplier ppm level at 10 and if possible decreasing it, increasing logistic performance and the last is obtaining price discounts on purchased goods.All of these strategies were assumed to be the most influential factors that would provide the organization sustainable leadership.
By using results of the Delphi method a pairwise linked FRM was developed for incorporating criteria with corporate strategies and then with sustainable leadership.
As it was mentioned in the first step, there were four decision makers in total.Therefore, there would have been four maps each of which were (111x7) matrix, representing relationship between criteria and corporate strategies, and another four maps each of which were (7x1) matrix, representing effects of strategies on SL.Moreover, 9 point scale of fuzzy linguistic variables would be used when evaluating the relationships between, criteria, strategies and competitiveness.Therefore, the size and the complexity of the questionnaire may have caused burnout in evaluators and required long time to be completed if four maps were required from the experts.Therefore, with consideration of reaching a consensus in the prior steps, only one map representing group evaluations was created instead of four maps.
 Step 7. Preparing the relational maps: In this step, two evaluation matrix were created in MS Excel as they can be partially seen in Figure 3 and Figure 4.The first one was a (111x7 matrix) representing the relationships between criteria and corporate strategies.The criteria were in the rows and corporate strategies were in the columns.The second one was a (7x1) matrix representing the relationships between corporate strategies and SL.The corporate strategies were in the rows and SL was in the column.Each cell was divided into sub cells that represented 9 point scale of fuzzy linguistic variables (Glykas, 2010) as listed in Table 4.The fuzzy membership functions representing fuzzy linguistic variables are given in Figure 5.This scale enable decision makers describe the grade of influence with linguistic variables like "strong", "weak" and etc.After the preparation was done, the spreadsheets was emailed to the experts for the evaluation.If there were one map for each evaluator, there would be one more step between Step 8 and Step 9.That would be aggregation of the maps or relational matrices.Suppose, there are two different answers for one relation, negatively medium and negatively weak.In this case, all fuzzy linguistic values of this relation would be aggregated into a graph as it is seen in Figure 8, and this step would be repeated for each relation that had more than one type of fuzzy linguistic value.Defuzzification is the conversion of a fuzzy quantity to a precise quantity which is called crisp value.The most commonly used method for the conversion is center of gravity (COG).In this method, center of the area below the (aggregated) membership functions is calculated.For the calculation, where crisp value was denoted by z and fuzzy membership function was denoted by μ_c ̃ (z), algebraic expression below was used (Alavala, 2008;Ross, 2010).The rest of defuzzified values of all fuzzy membership functions were calculated with the same method and the results are given in Table 5.

Results of the criteria determination part
Results given in this section cover the first five steps of the process, which are about criteria determination.General findings of this part are as follows: Delphi rounds were made at the beginning of summer since business was much steady in this time of the year comparing to other seasons.The first and second rounds took 18 and 34 days to be completed and some details about the experts can be seen in Table 6.As it was explained in Step 2, a list of possible criteria regarding the supplier selection context was prepared with the procurement director in order to reduce the number of Delphi rounds and prevent burnout of the experts.In the prepared list, there were 106 criteria.At the end of the Delphi rounds 111 criteria were obtained.Even though, the number of the criteria seemed to be increased after the Delphi rounds, there were not actually any new criteria proposal but propose of using same three criteria under different headings as listed in Table 7.The repeated criteria were flexibility, ability to fulfill unexpected orders and commitment to quality.Flexibility and ability to fulfill unexpected orders were both used in technical capabilities of human resources and also in technical capabilities of production system.Commitment to quality was used both in product quality and system quality.Moreover, organizational structure and length of contract criteria were divided into two complementary criteria.The first one was divided into family owned business and corporate organization criteria.The second one was divided into short term and long term contracts criteria.Therefore, there were 111 criteria in total to be related with the corporate strategies of the company.

Group name
No Criteria The biggest portion(is hold by main customer) in the supplier's budget C111 Capability to support my company in having competitive advantage Table 7. Final criteria list * Repeated criteria **Please see the Appendix for the explanation At the beginning of Delphi rounds, there were six experts in total.However, two of them left the company between the first and second round.One important finding was that, when there were six evaluators, variance of two criteria was zero.They were ISO 9001 and product quality.However, when two of them left the company, number of zero variance criteria increased to twelve.With their rates, they were ability to use computer technologies (3), able to meet JIT conditions (4), number of stopping production line (5), product quality (5), ISO:9001 (5), ISO:16949 (4), existence of continuous improvement systems (4), geographical location (4), recall insurance (4), environmental awareness (3), social responsibility (3) and collaboration with universities (3).Additionally in the first round, variance of three criteria was equal and greater than one.They were continuous employee training (σ= 1), ability to speak same language (σ= 1,09) and supplier's supplier ppm (σ= 1,225).
According to second round evaluations, some other criteria had zero variance.They were previous production level (3), amount (%) dedicated to training resources (3), having proper machineries for quality controlling (4), commitment to quality (4), internal ppm (3), inventory capacity (4), sales income distribution ratio (4), after sales support (4), eagerness for cooperation (5) and capability to support my company in having competitive advantage (4).Criteria of which variance were equal and greater than one in the first round, had variance lower than one in the second round.

Results of the influence estimation part
Results given in this section cover the last six steps of the process which are about the influence estimation, Fuzzy relational maps technique was used for the estimation.With this technique, effects of each criteria on SL through corporate strategies was obtained as listed in Table 8.As it was stated in Step 11, criteria having values equal and greater than (+3) or equal and smaller than (-3) assumed to be strategic criteria.There were 27 criteria assumed to be strategic.However, none of them is negative.Even though, there were six criteria of which values were smaller than zero, value of none of them is equal and smaller than (-3).However, these negative criteria may be considered critical when pre-qualifying candidate suppliers.Also, there were some criteria of which values were in the interval (-1, 1).These criteria may be considered as unimportant in strategic supplier pre-qualification context.A list of strategic, critical and unimportant criteria and criteria-strategy-sustainable leadership map is in Table 9 and Figure 11.In this research study, Delphi method was used for only collecting names of the possible criteria that was considered important regarding supplier selection context.As there were only four evaluators, there was not enough sample data to make some statistical calculations with.Therefore, criteria values obtained in Delphi rounds were not considered to be important to be discussed about in this section.However, there are some interesting findings about the Delphi rounds.The first finding is the length of the rounds.As it is stated in Section 5.1., the second round took two times longer than the first round.The reason might have been that experts were asked to reexamine their first round answers.Also the second Delphi round was hold in the middle of summer, which was vacation time.Therefore, it can be assumed that the time of the year is an important issue for timing of a research project that has Delphi method application.
Another finding is about the experts.As it can be seen in Table 7, there is one one year experience (D) expert and one fourteen year experience (B) expert.Normally, it is expected that the fourteen year experience expert has less variance in his/her answers between rounds as he/she is much more experienced than D. However, expert B did more changes in his/her evaluations than expert D did.Also four of his/her evaluations left out of the range at the end of the rounds while all of the answers of expert B was in range.The situation of expert D may be explained with the person-organization fit theory, which is defined as the congruence between values of employee and values and norms of organizations.Also, as explained in Step 1 in Section 4, two decision makers resigned between the Delphi rounds.With the resignation, number of zero variance criteria was increased from two to twelve in the first Delphi round.This increase also may be explained with the congruency theory.Values, norms and characteristics of the resigned employees may not be compatible with the organization's.In other words, fitness between employees and organization may be low.Low person-organization fit, especially in procurement department of an organization, may result in disastrous procurement decisions and business partnerships since these employees would make procurement decisions according to their own values not the values of the organization.
As it was stated in Section 5, variance of all criteria was lower than one, and 22 criteria had zero variance at the end of the Delphi rounds.That is more than 20% of all criteria.Therefore, it can be concluded that experts reached a consensus with the Delphi rounds that provide a common point of view about the criteria.
The very most interesting finding about the outcomes of FRM approach was that price and cost were not in the list of strategic criteria.Their influence on SL of the company through corporate strategies were estimated moderate.However, they were two of the most important evaluation factors of the company.Also they are considered important in most of the studies in the literature (Cao et al., 2014;C. G. Şen et al., 2010;Sarkar & Mohapatra, 2006).However, there is a difference in practice of this research.Importance of all criteria were estimated in strategy and leadership context of the company.Having this result actually what was expected.As it was stated previously, this research study aims to define a set of strategy based criteria that help to eliminate strategically misfit companies in the pre-qualification step of Alphanumeric Journal Volume 4, Issue 2, 2016 supplier selection process.Cost and price cannot be determined before the final selection because this information can be obtained when quotation request is sent to candidate companies, and usually it is sent just before the final selection.Therefore, these criteria cannot be considered as pre-qualification criteria.As a result, if experts in procurement departments make strategy driven procurement decisions, their final choice may change since the set of considered criteria get changed.
Another interesting finding is about the top five criteria.Existence of continuous improvement systems was the first, technical qualification of the employees and external communication ability were the second, business strategy of the candidate supplier was the third, experience of the company was the fourth and ISO:16949 was the fifth important strategic criteria.As it is clear, organizational and managerial characteristics of the company takes the top five space in the strategic criteria list.Therefore, organizations need to move their attention from monetary / financial criteria to much more managerial criteria when selecting supply chain partners if they want to obtain sustainable leadership in the market.For the supplying company side it is the same.Supplier companies need to invest more in their management systems so their human resources if they want to be a business partner of global companies.Moreover, to the best of our knowledge, this kind of criteria are not usually considered in real life practices.Integrating these criteria into supplier selection and evaluation system of organizations may cause change in procurement decisions of experts.
Number of stopping production line, having short term contract, eagerness of sharing confidential information and being family owned business had negative effects in strategic context on the leadership of the organization.Since their absolute influence was not as high as strategic criteria, they were grouped as critical criteria that may be considered in the final selection of suppliers.Moreover, there is one important point to pay attention on.Having/preferring short term contracts with business partners was considered like a defect by the subject company.In other words, subject company prefers suppliers that have long and most probably strong relationships with its partners.This finding is in accordance with the related literature.
Another interesting finding was about unimportant criteria.For example, influence value of social responsibility and environmental awareness are estimated very low, close to zero.However, these two factors are most popular research and investment area of today's business market.Therefore, a further analysis, for example a more expansive industry based research can be done about this finding.
The list of strategy based supplier pre-qualification criteria determined with this research can be divided into two sets of criteria.While the first set includes criteria about organizational texture of candidate companies, the second set may include technical criteria.This structure enables decision makers apply a stepwise procedure of supplier pre-qualification model.By doing so subject company may eliminate organizationally misfit nominated candidate suppliers with the first set of criteria and then classify the rest with the second set of criteria according to their technical capabilities.As a result, nominated candidate suppliers that pass the organizational evaluation and have high score in technical evaluation become candidate supplier for the final selection.Regarding the methods used for modelling this framework, Delphi technique was used for only extracting knowledge about important selection criteria and reaching consensus about them.Other techniques like brainstorming was not used since this kind of group decision making techniques inherit some pitfalls.Moreover, only the experts, who had direct relationship with suppliers, involved in this research study.Instead, decision makers, who indirectly affect procurement decisions, from different units of the organization like production or quality departments may also involve in Delphi rounds.Also in this research, a large set of item was determined and evaluated.In a future research, Delphi technique may be used to estimate importance values of only strategy based criteria obtained in this study.
Fuzzy relational maps technique was used in estimating the importance of strategy based supplier pre-qualification criteria for the goal of the company which was sustainable leadership.In classical FRM approach, one map is created for each expert.However, since there was many relationships to be considered, only one map was created for all experts.The small set of strategy based criteria determined in this research can be evaluated by each expert individually in a future research as well.Also in this research, interrelationship among criteria was not investigated.Therefore, fuzzy cognitive map approach can be applied to this kind of investigation in a future research.
To summarize, the supplier selection is one of the most important tasks for organizations since results of the final decision affect profitability, position and market share of the company (Aghai et al., 2014;Cheraghi et al., 2011;Kasirian & Yusuff, 2013;Lee, 2009;Lin, 2012;Xiao et al., 2012).Even though there are voluminous work about this subject in the literature it is still a problematic area as many conflicting subjective criteria and numerous candidates are involved in the selection.This makes the process complicated.To ease the final selection and to find the most suitable supplier, it is essential to define the right criteria and apply a prequalification step that shortlists the number of candidate suppliers.Also, strategic fitness is another issue to be considered in supplier selection context.Supply chain systems require a synchronous operations between organizations and it can be achieved only between strategically congruent organizations.Therefore, it becomes important to look for strategic fitness in candidate suppliers.Therefore in this research study a hybrid approach that incorporates Delphi technique with fuzzy relational maps was proposed for determining strategy based supplier prequalification criteria and their influence on sustainable leadership.
Alphanumeric Journal Volume 4, Issue 2, 2016 Appendixes Explanations of Some Pre-evaluation Criteria  "Business strategy" represents four types of organizations, follower or reactor, analyzer, defender and prospector or innovator.
 "Internal communication ability" represents effective usage of communication means within the candidate company.For example, usage of e-mail address, phone number, information boards, Andon boards, employee suggestion system, company bulletins or newspapers and social activities, etc.
 "Commitment to quality" represents allocating a budget for corrective & preventive actions, quality trainings and internal audit systems, tracking quality costs, having periodic review meetings and encouraging attendance of employees to these meetings.
 "Eagerness to cooperate" represents that the candidate company is cooperative, communicative and negotiable in case of any problem.
 "Strategic importance for my company" represents that the candidate company is very valuable with respect to its core competencies, technology and patents and therefore it may foster the market position of the buying company if worked with it.It is also strategically important if the candidate company is a monopoly.
 "Technical qualifications of employees" represents that blue and white collar employees of the candidate company are experienced in their fields; their educational background is related to their work; they have comprehensive knowledge about their work.
 "Quality control" represents that the candidate company effectively use statistical process control, poke yoke and quality gate tools.
Figure 1 illustrates a FCM of which each node Ci represents concept variable and each arc wij represents connection weight (effect of Ci over Cj) in the interval [-1, 1].Cycled feedback graph structure of FCMs allows what-if inferencing.

Figure 1 .
Figure 1.Fuzzy cognitive map Figure 2 illustrates a typical FRM.Range space is denoted by Rm and domain space is denoted by Dn.Connection weights that show the relationship between domain and range nodes are denoted by wij, in the interval [-1, 1].

Figure 2 .
Figure 2. Fuzzy relational map C, D and R.They have m, n and t sets of nodes in the spaces.Directed graphs relating C and D and relating D and R can be inter-related indirectly.In other words FRMs enable users to define and estimate the hidden link between indirectly related spaces.If W1 is the connection matrix relates C and D, then W1 is a m x n matrix.The product of the connection matrices belongs to C and D is called the hidden connection matrix, and the map (graph) represent this relation is called the hidden directed graph of the pairwise linked FRMs.If W2 is the connection matrix relating D and R, then W2 is an n x t matrix.Therefore, W1W2 is an m x t matrix which is the connection matrix relates C and R. Also W2 T W1 T matrix relates R and C. Fuzzy relation between concepts means fuzzy causality that may have various degrees of membership representing strengths of the relationship between the concepts.It may have values in the interval [-1, 1].The relation may express three different situation; positive causality where wij>0, negative causality where wij<0 and no relationship where wij=0.Please see an algorithm of Groumpos (2010) for Alphanumeric Journal Volume 4, Issue 2, 2016

Figure 3 .
Figure 3.A part of the first relational matrix

Figure 4 .
Figure 4.A view of the second relational matrix

Figure 5 .
Figure 5. Fuzzy membership function  Step 8. Fuzzifying the relations: After all evaluations were done by the experts, values in the matrix, in other words, influences between nodes, were fuzzified.A sample part of the evaluation and fuzzification of the matrix can be seen in Figure 6 and Figure 7.

Figure 6 .
Figure 6.A part of the relational values of the first matrix

Figure 8 .
Figure 8.An aggregated graph of NM and NW fuzzy membership functions  Step 9. Defuzzifying the relations: After fuzzification of the relationships, their crisp values were calculated based on the fuzzy membership functions given in Figure 5.Defuzzification is the conversion of a fuzzy quantity to a precise quantity which is called crisp value.The most commonly used method for the conversion is center of gravity (COG).In this method, center of the area below the (aggregated) membership functions is calculated.For the calculation, where crisp value was denoted by z and fuzzy membership function was denoted by μ_c ̃ (z), algebraic expression below was used(Alavala, 2008;Ross, 2010).


If there were two different answers for a relation as in Figure8, similar calculations would be done as below.Line equation of the first line segment between (-0.75, 0) and (-0.5, 1): +0.75 0.25 Line equation of the second line segment between (-0.5, 1) and (-0.25, 0): (−−0.25)Step 10.Creating crisp matrix: After defuzzifying fuzzy relations, both fuzzy relational matrix representing the relationships between "Criteria & Corporate strategies" and "Corporate strategies & SL" redesigned with crisp values as it is seen in Figure 10.

Figure 10 .
Figure 10.A part of the deffuzified first matrix  Step11.Obtaining final results: Eventually, product of these two matrix was the effect of each corporate strategy based criteria on the SL.MATLAB R2014a was used to obtain the product of two matrix.Based on the results, the criteria of which crisp values were equal and greater than (+3) or equal and smaller than (-3) assumed to be strategic criteria.

Figure
Figure 11.Criteria-Strategy-Leadership map

Table 1 .
Supplier pre-qualification criteria used in previous studies

Table 2 .
Techniques used for determining supplier pre-qualification criteria in previous researches 2.4.Strategy based supplier selection / pre-qualification According toKar & Pani (2014), nowadays senior management of organizations consider procurement as a key factor in strategy formulation.However the literature pertaining to supplier selection model fails to appreciate strategic procurement models (Surajit

Table 8 .
Influence of each criterion on sustainable leadership

Table 9 .
A list of strategic, critical and unimportant criteria