Phase 1. Definition of criteria and key indicators used in the supplier evaluation process

The results of this phase guarantee that the improvement in competitiveness will create value to buyers because any action is focused on the upgrading of critical evaluation criteria. This phase comprises two stages:

The identification of criteria and sub-criteria is based on theoretical and empirical basis. Preliminary criteria are defined in terms of the existing literature, and then validated through a discussion with a panel of experts integrated by critical decision makers from the automotive sector. The phase is completed by the operationalization of each criteria through a meaningful set of tangible key indicators validated by the experts. For example the key indicators defined to measure the criterion “product and process quality” are: number of rejected parts, PPM (defective parts per million of parts produced), downtime hours, quality certifications obtained by the supplier, failure contention and traceability systems, quality control of incoming materials, and training and certification of suppliers.

After identifying key criteria and sub-criteria, the panel of experts is required to assign weights of importance to each one. The quantitative technique used to accomplish this task is Fuzzy Analytical Hierarchy Process (FAHP) which is an extension of the well-known AHP methodology that considers the vagueness of the linguistic variable describing the importance of a criteria. It is assumed decision makers clearly understand the meanings or the importance of linguistic values but not everyone has the same perception of a particular description. For example, the term “definitely more important” may be equivalent to a noticeable difference in the mind of a DM but interpreted as a strong difference by another. FAHP captures this ambiguity by using fuzzy sets and membership function to assign quantitative rates to the DM pairwise comparisons between criteria at the same hierarchical level.

The consistency of the individual opinions is assessed before combining them to get a final set of weights for each criteria and sub-criteria in the evaluation hierarchy. Following FAHP, suppliers are evaluated on each criterion and the corresponding weights used to compute a compensatory global score of performance or competitiveness. But this score is too aggregated to allow the identification of suppliers with good chance for development, therefore we suggested the construction of maps of relative competitiveness.

Phase 2. Construction of a conceptual map of competitiveness for suppliers

The conceptual maps depict graphically the relative level of performance (competitiveness) of a group of suppliers. The input of the map are the supplier’s evaluation performed by DM according to the criteria and sub-criteria defined in the previous phase. Each supplier is rated on a scale going from 0 to 10, where 0 means the lowest acceptable level in a key indicator and 10 represents the ideal level. The scores of all indicators belonging to a particular criteria are averaged and then weighted according with the relative importance of each criteria and sub-criteria. Finally each supplier is represented in a 2- or 3-dimensional map thanks to the use of Multidimensional Scaling (MDS) techniques.

The graphical output of MDS reflects the appreciation of decision makers regarding the relative competitive position of each supplier. In the map, the Euclidean distance between suppliers represents similarities (or dissimilarities) between them, and the distance with respect to the “ideal supplier” shows the “lack of competitiveness” of a particular supplier. Clusters or groups of suppliers with similar levels of competitiveness are easily identified. Suppliers in the same cluster have equivalent levels of global performance or competitiveness, meanwhile groups of suppliers with very different levels of performance are placed far away from each other on the map.

The visual representation of the complex set of relationship among suppliers is simplified by MDS. The underlying dimensions “explain” the similarity between suppliers. For example, two suppliers may be close (similar) because they have similar rates of performance in terms of hard criteria like quality, price and on-time delivery. Thus, the similarity between suppliers is a function of their similarity in scores across a particular subset of attributes representing “global key” or “substantive” indicators for decision makers. These global key attributes are a simplified aggregation of individual criteria that decision makers within an industrial sector ultimately use to distinguish one supplier from another.

It is important to realize that these substantive dimensions or simplified keys are expected to be considerably smaller than the number of criteria used to rate suppliers. This is because the performance criteria, while cognitively distinct, may be highly inter-correlated in the minds of experts (e.g. quality, on-time delivery and clear contingency plans jointly describe a dependable supplier or be associated with tangible results) and therefore contain some redundant information. Then the basic utility of the perceptual map is the identification of:

Once suppliers judged suitable for development are identified, the particular contents of a SD program need to be specified during a last phase.

Phase 3. Outline a customized supplier development program that increases the competitiveness of a supplier and its opportunity of being part of the automotive supplier chain

Beginning with phase 3, an “acceptable” cut-off level of global performance was defined by the DM after using a nominal management technique during the workshop. ^{Cooper (2007)} recommends setting this lower bound as the average of all performance rates, but depending on the strategic importance of the product supplied, the number of suppliers and the resources available to implement the SD program, other bounds may be defined. In this proposal, the maximum performance level was defined in terms of the supplier with the best observed performance provided this supplier got a minimum “9” score on all first-level criteria, otherwise the maximum corresponded to the “ideal” supplier (a score of 10 on all criteria). The normalized supplier’s performance ratings were used to compute inefficiency gaps. By following the Pareto principle, those criteria that account for roughly 80% of the total weights of importance were analyzed during the following phase to define proper activities for development.

The areas where the supplier achieve a performance level equal to or greater than the acceptable cut-off level were judged “satisfactorily efficient” and may not require immediate support while those areas where the supplier’s performance gaps are the largest should be subject to a problem’s root causes analysis (^{Weber et al., 1998}). This means experts need to identify the causes of underperformance by checking on the specific indicators and suggest a set of customized development activities. The time and resources required to attain an acceptable level of performance was estimated from the opinions of DMs and academics from a major Mexican university with large experience in training programs and improvement projects involving companies in the automotive sector.

Finally, specific activities were assigned to suppliers by applying a third quantitative technique, Goal Programming (GP). This technique was selected because it helps to solve problems with multiple goals or objectives (sometimes in conflict) when is not possible to find a solution that optimizes all objectives. We assumed there are restrictions in time and budget to any development program because of the limited resources available to complete the project. Therefore, the problem was defined as what activities assign to each supplier to get the maximum improvement in performance for all participant suppliers given the restrictions on resources. The main idea of GP is to transform the multiple criteria of performance (price, quality, accuracy of delivery, etc.) on a single goal (global competitiveness) and get an efficient solution to this problem even though it is not optimal with respect to all the objectives. This implies maximizing the global efficiency of the suppliers’ base by selecting those projects that minimize the performance gaps in the different areas -and consequently increase the global competitiveness - while considering the limited available budget and the specific time frame of the SD project.

The results of applying the described methodology are discussed in the next section.

Phase 1

The hierarchical structure representing the criteria used to evaluate and select suppliers in the automotive sector was defined through a literature review and a workshop with experts from lead firms in this sector -Gates Rubber, Parker Fluid Connectors, ZF Lemforder, Robert Bosch, Trelleborg, Gestamp, Mahle de México, Maccsa-.

Figure 2 shows the resulting hierarchy. The degree of importance of each criteria and sub-criteria was computed with Fuzzy AHP. Table 1 and Figure 3 show the global weights of importance for the first level criteria. The most important criteria were in total agreement with the literature review and include: purchase price, product & process quality and reliability of delivery. Interestingly, criteria like “risk management” had one of the lowest weights while criteria such as “greenness” were not included in the hierarchy. In the case of “risk management”, the low importance was attributed to the fact current suppliers have capacity in excess and are located nearby the BF, therefore no interruptions of supplies are expected. Green criteria on the other hand are not currently used because existing environmental regulations are relatively easy to fulfill.

Figure 2 Hierarchical structure of the criteria used in the supplier evaluation process 

Figure 3 Global weights of first level criteria 

General scores of supplier’s performance can be computed at this phase to rank suppliers but they are insufficient to define meaningful SD activities and visualize the effort required to improve current performance on the defined criteria. Phases 2 and 3 are intended to overcome these difficulties.

Phase 2

Following the proposed methodology, the next step was to build a perceptual map to picture suppliers according to their levels of competitiveness. The perceptual maps constructed by applying MDS (Figure 4) graphically describe how competitive are nine of the strategic suppliers of the automotive sector while Table 2 shows the coordinates of each supplier in the map along with the distance to the ideal supplier.

Figure 4 Map of Euclidean distances between suppliers automotive sector 

Based on the map, different approaches may be applied to perform the identification of viable candidates for development; benchmarking current suppliers versus world-class suppliers and performing a Pareto analysis are two recommendable choices (^{Krause et al., 1998}; ^{Hahn et al., 1990}). Under the benchmark approach used by American auto companies, suppliers are classified into four ordered classes. Suppliers with a world-class performance are assigned to the high-level class; these suppliers do not require more development efforts than feedback and incentives. In the second class there are suppliers with small shortages on only few core criteria. Therefore these suppliers have the potential to achieve the required performance levels and are the most viable candidates for development. Suppliers with important performance deficits on several core criteria are assigned to the third-class; these suppliers need to reduce their performance gaps before they are judged viable candidates for development. Finally, the suppliers in the lowest class have limited capacity or exhibit significant deficiencies in multiple criteria, then they are scheduled for elimination (^{Hahn et al., 1990}).

Following the benchmark approach, concentric circles around the “ideal supplier” were drawn in the MDS map in order to identify the most competitive suppliers. At least one supplier is included in each circle and all equidistant suppliers (Euclidean distance) to the “ideal supplier” have the same level of global competitiveness even when their performance on specific criteria could be different.

At Figure 4 and Table 2 we can observe that supplier 3 has the highest level of competitiveness (0.930) because is the closest to “ideal supplier”; in contrast supplier 4 is the least competitive (0.649). Groups or clusters of suppliers with similar level of competitiveness are suppliers {7, 9} and suppliers {1, 2, 6}. Results are used to decide which group of suppliers includes the more suitable candidates for development. BF may establish a minimum level of competitiveness to admit suppliers in a SD program. For example assume suppliers with a level of competitiveness greater than 0.850 are judged highly competitive to improve their performance by themselves; suppliers with a level of competitiveness between 0.700 and 0.850 are suitable candidates for development while suppliers with a level of competitiveness below 0.700 require too much investment to be developed. Then suppliers 3, 7 and 9 will only receive feedback and incentives, suppliers {1, 2, 5, 6, 8, 9} would be considered viable SD candidates while suppliers {4} are not considered for development and can be discarded from the supplier base unless the capacity of the other suppliers is insufficient to fulfill BF demand.

In case new suppliers are available or the initially discarded ones improve their performance to the acceptable levels, they would be included in the perceptual map to determine if they are viable candidates for development.

Phase 3

This last phase defines the areas to be developed on each supplier based on the analysis of the reasons that explain their positions on the map. In order to show how it works we take the case of the suppliers 1 and 2.

According to the previous results, these suppliers have very similar levels of global competitiveness (0.774, 0.778) but not so far from the ideal, therefore they are eligible for development. The areas of improvement correspond to the largest gap with respect to the ideal performance, to help experts to visualize the situation, differential gaps were represented in Figure 5. The dashed line in the graph corresponds to the lowest acceptable level of performance (efficiency index of 0.85). The areas or criteria where the supplier achieve an efficiency index equal to or greater than this bound are considered “satisfactorily efficient” and do not require improvement while areas where supplier has an efficiency index below this level must be considered eligible for root cause analysis and posterior development. Following the steps suggested by ^{Hahn et al. (1990)} regarding the design of a SD program, a panel of experts identified the “root causes” of underperformance of key indicators of each criteria (Table 3) and suggest appropriate and customized development activities (projects).

Figure 5 Normalized performance levels of suppliers 1 and 2 in the most important criteria 

The areas to be taken into account in the root causes analysis are the following:

Supplier 1: Purchase price, product and process quality and reliability of delivery.

Supplier 2: Product and process quality, after-sales service and organizational culture.

SD projects for each of the areas where a supplier did not reach an acceptable level of performance were defined in terms of the practical knowledge of the group of selected experts. The time and cost of implementation of each project as well as the expected improvements in performance were estimated by the panel of experts and are shown in Table 4. The total time of implementation is computed by assuming projects or activities are implemented sequentially because there is a restricted scheduled budget.

Usually there are restrictions on time and budget to be spent in development programs. In this case, priority activities were chosen through the solution of a GP model. The decision variables X_ijk are the activities to be implemented in order to achieve the largest improvement in performance. The sub-indexes identify the supplier (i), the area or criteria requiring improvement (j) and the number of projects or activities defined for this area (k). Each project takes a time to be completed (t_ijk), involves a cost (c_ijk) and results in an improvement in performance (a_ijk).

The aim is to maximize the efficiency of suppliers by selecting those projects that minimize the performance gaps in the different areas (b_ij) and consequently increase the global competitiveness of suppliers while considering that the buyer has a limited available budget (M) and needs to develop the supplier during a specific time frame (T).

The linear programming formulation is as follows

Minimize

Subject to

For i=1,…,m      j=1,…,n      k=1,…,r

The development activities assigned to each supplier after solving the GP problem given a maximum authorized budget of 100,000 USD to develop each supplier during a year (12 months) are shown in Table 5.

Once selected projects are implemented, the suppliers’ performance is expected to change after a year. Figure 6 depicts the expected reductions on the performance gaps: the level of global competitiveness of supplier 1 would improve from 0.77 to 0.86 while the global competitiveness of supplier 2 would increase from 0.78 to 0.89. These new levels of competitiveness would allocate suppliers to the “satisfactory performance” class (above the lower bound of 0.85 defined in the previous stage).

Figure 6 Expected performance gaps after development of suppliers