4. Examples of Supply Chain Decisions Trading Off Criteria

This study focused on a method to identify suppliers with emphasis on sustainable awareness. They utilized the triple bottom line paradigm, considering sustainability along with social and financial aspects. These authors used TOPSIS, ² which has the relative advantage over SMART of being applicable to large numbers of alternatives. But the treatment of criteria, weights, and scores is common to TOPSIS and SMART. They also used fuzzy modeling, and a group context, but since they converted fuzzy input to crisp numbers and used a group decision set of preferences, the method works the same once numbers are modified. Based on the data provided in their paper, we can infer the following risk order. We provided specific swing weights in Table 4.1:

Scores for suppliers (which we made up) on each criterion are given below, along with resultant value scores in Table 4.2:

Supplier 1 has stronger characteristics with respect to quality, environmental issues and safety, but is expensive. Supplier 2 has good quality, moderate risk and moderate cost. Supplier 3 has lower quality, higher risk, and lowest cost. With this set of scores and one particular set of weights, supplier 2, who emphasizes quality, has the highest score. Value analysis provides a means to utilize these scores to identify areas of potential improvement.

This study applied fuzzy AHP and TOPSIS to a model to assess risk for a supply chain. Thus it did not involve a specific decision, but rather was meant to provide a metric for evaluation of overall supply chain risk. They had four top-level categories of risk {supply risk, demand risk, process risk, and environmental risk by which they included political environment}. The methodology, as in Case 1, could be applied to evaluation of suppliers as well. Table 4.3 gives criteria and shows weight generation, using Samvedi et al. general criteria ranking and our own specific values.

Table 4.4 shows scores for three suppliers the might be evaluated. Supplier 1 might be a high quality, high cost alternative, Supplier 2 a bit inferior to Supplier 1 on cost but higher on quality, and Supplier 3 located in a higher risk area.

The value score can be used to rank suppliers. Here Supplier 2 is better than Supplier 1, and both are radically better than Supplier 3.

This study used TOPSIS and intuitionist fuzzy multi-criteria decision making modeling to evaluate alternative vehicle technologies. They compared seven types of vehicles with 16 criteria (using the triple bottom line paradigm of economic, social, and environmental) as in Table 4.5:

The seven vehicle types were:

Table 4.6 gives scores and shows value calculations:

This study used ELECTRE multi-criteria decision making modeling to evaluate oil and gas companies, again using the triple bottom line. They compared the biggest five global oil and gas companies with two economic, ten environmental, and three social criteria as in Table 4.7:

Infante et al. evaluated firms over time, with scores provided for each year from 2005 through 2010. We will base our scores to reflect 2010 numbers in their data. The matrix of scores for each criterion by option are given below, along with calculation of overall value score. Table 4.8 shows input measures from the original article:

Infante et al. utilized equal weights, and then demonstrated sensitivity to weights in some variants. Based on ELECTRE approaches, scores were generated in one of the metrics that method offers, with a score of 0 below some minimum (a q parameter) and 1 at or above some maximum (a p parameter). The scores in Table 4.9 reflect a linear formulation for input measures between q and p.

Infante et al. utilized equal weights, with weight changes representing sensitivity analysis. Table 4.10 shows results for various combinations of weights. The first set of weights divides 1/3rd by the number of criteria within each category, yielding weights of 0.167 for economic factors, 0.033 for environmental factors, and 0.111 for social factors. The second row assigns each of the 15 criteria a weight of 0.067, which is equal for all, but biases analysis toward environmental factors because there are ten measures as opposed to two or three. The last three rows show relative emphasis on economic, environmental, and social factors in turn. Highest value function for each oil company is identified in bold:

When economic factors are emphasized, Exxon Mobil performed well in 2011. When environmental factors received heaviest weight, BP did best (possibly in response to Gulf of Mexico oil spill history). BP also did well when social factors were emphasized, again possibly explicable in light of recent history. Infante et al. did a commendable job in looking at annual performances. Here our point is to demonstrate use of multiple criteria models, in this case as a performance measure.

This application used AHP in the style of a business scorecard. While the authors gave a good discussion of criteria and its components, the data they provide for relative weights referred only to the top level factors of On-time delivery, completeness, correctness, and damage/defect free products. They also gave examples demonstrating scoring of departments within the organization on each of these four criteria by managerial subjective assessment, as well as using a more metric-driven model. Furthermore, they gave ranges for relative weight importance (which could be used for alternative multicriteria models ⁷ such as HIPRE). ⁸

In this study, data for relative criteria importance was given in ranges. We present the extremes below in development of SMART weights in Table 4.11:

The last two criteria have fairly consistent weights, so we chose weight of 0.21 for Correctness and 0.14 for Damage-defect free products. The first two had quite a range, as each extreme had a different first selection. Using the maximum weight for the first and subtracting 0.35 as the weight for the third and fourth ranked criteria, weights were generated. Using on-time delivery as the most important criteria yielded a weight for completeness outside the extreme weights, so we raised that weight to 0.29, lowering the weight for on-time delivery to 0.36. No adjustment was necessary to keep weights within range for the set of weights assigning completeness the greatest weight as shown in Table 4.12:

Gaudenzi and Borghesi gave two sets of scores to evaluate risks within their departments. Scores based on managerial input as well as a model used by Gaudenzi and Borghesi are demonstrated with both sets of weights generated above. Scores here are presented in a positivist perspective, with 1.0 representing the best performance. Therefore low resulting scores are associated with the most problematic departments.

The first set of value scores reflect weights emphasizing on-time delivery, with manager subjective scores shown in Table 4.13:

The scores themselves highlight where risks exist (0 indicates high risk, 0.5 medium level risk). The value scores give something that could be used to assess overall relative performance by department. Order cycle has no problems, so it has to score best. Manufacturing seems to have their risks well under control. Procurement and transportation departments are more troublesome.

The second set uses the same weights, but scores based on model inputs (see Table 4.14):

The implications are similar, except that the warehousing department shows up as facing much more risk.

We can repeat the analysis using weights emphasizing completeness. Using managerial subjective scores (Table 4.15):

This set of weights gives the transport department a better performance rating, but otherwise similar performance to the earlier analysis.

Finally, we use the model scores for weights emphasizing completeness (Table 4.16):

Here the warehouse department appears to face the greatest risk, followed by the procurement department.

The Guadenzi and Borghesi article presents an interesting application of multiple criteria analysis to something akin to business scorecard analysis, extending it to provide a potential departmental assessment of relative degree of risk faced.

The cases presented here all applied multiple criteria models. This type of model provides a very good framework to describe specific aspects of risk and to assess where they exist, as well as considering their relative performance. The value scores might be usable as a means to select preferred alternatives or as performance metrics. Through value analysis, they can direct attention to features that call for the greatest improvement.

Value analysis can provide useful support to decision making by first focusing on hierarchical development. In all five cases presented here, this was accomplished in the original articles. Nonetheless, it is important to consider over-arching objectives, as well as means objectives in light of over-arching objective accomplishment.

Two aspects of value analysis should be considered. First, if scores on available alternatives are equivalent on a specific criterion, this criterion will not matter for this set of alternatives. However, it may matter if new alternatives are added, or existing alternatives improved. Second, a benefit of value analysis is improvement of existing alternatives. The score matrix provides useful comparisons of relative alternative performance. If decision makers are not satisfied with existing alternatives, they might seek additional choices through expanding their search or designing them. The criteria with the greatest weights might provide an area of search, and the ideal scores provide a design standard.