Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

Vendor Selection Using Principal

Component Analysis

AUTHORS
INTRODUCTION
Alberto Petroni During recent years, supply chain management and
is a researcher in the Department of Industrial Engineering at the the supplier (vendor) selection process have received
University of Parma in Parma, Italy. He earned his Ph.D. degree at considerable attention in the business management
literature. In one of the widely cited articles on opera-
the University of Padua in Italy. Dr. Petroni’s research interests tions strategy, Miller (1981) classifies supply strategies
include operations management, supply chain management, and as one of the strategic operating choices.
technology management. Additionally, with the increase in use of total quality
management (TQM) and Just-In-Time (JIT) concepts by
Marcello Braglia a wide range of firms, the supplier selection question
is an assistant professor at the University of Brescia in Italy. He earned has become extremely important. In a recent study,
his M.Sc. degree from the Politecnico di Milano. Mr. Braglia’s research Flynn et al. (1994) found supplier involvement to be
an important dimension of quality management. Banker
interests include combinatorial problems in manufacturing systems, and Khosla (1995) classify the supplier selection process
maintenance, and logistics. as an important operations management (OM) decision
area. They suggest that OM research should attempt to
identify the supply chain management practices that
Purchasing managers need to periodically evaluate provide competitive advantage. Karmarker (1996) also
supplier performance in order to retain those sup- identifies supply chains as multidisciplinary in nature
pliers which meet their requirements in terms of and recommends an integrated OM-marketing
several performance criteria. The evaluation element approach.
typically consists of identifying the attributes, criteria,
Dickson (1966), in one of the early works on supplier
or factors relevant to the decision and then measuring
selection, identified over 20 supplier attributes which
or rating each vendor by considering each of the rele-
managers trade off when choosing a supplier. Since
vant factors. A critical part of the
then, a number of conceptual and empirical articles
SUMMARY overall supplier selection process
on supplier selection have appeared (for an exhaustive
is the determination of the relative
review, see Weber et al. 1991). The conceptual articles
importance of each of the factors. This article pre-
by Ansari and Modarress (1980; 1986), Benton and
sents an alternative decision model to evaluate the
Krajeski (1990), Bernard (1989), Browning et al. (1983),
relative performance of suppliers that have multiple
Burton (1988), Hahn et al. (1983), Jackson (1983),
outputs and inputs. This approach is based on a
Kraljic (1983), Sheth (1973), and Treleven (1987) are
multivariate statistical method, principal compo-
examples of publications emphasizing the strategic
nent analysis, that uses information obtained from
importance of the supplier selection process. The above
eigenvalues to combine different ratio measures
articles (and several others) highlight the trade-off among
defined by every input and every output. The
quality, cost, and delivery performance measures in the
method has been employed to aggregate multiple
supplier selection process.
performance measures for a real-world data set of
suppliers of a medium-sized firm operating in the A number of empirical articles on supplier selection
The Journal of Supply Chain
bottling machinery industry. have also appeared. Based on empirical data collected
Management: A Global
Review of Purchasing from 170 purchasing managers, members of the National
and Supply Copyright Association of Purchasing Management, Dickson (1966)
© May 2000, by the National
Association of Purchasing identified quality, cost, and delivery performance history
Management, Inc. as the three most important criteria in vendor selection.
Cardozo and Cagley (1971), Chapman (1993), Chapman

The Journal of Supply Chain Management | Spring 2000 63


Vendor Selection Using Principal Component Analysis

and Carter (1990), Dempsey (1978), Hakansson and supplier performance evaluation: categorical, weighted
Wootz (1975), Monczka et al. (1981), Wagner et al. (1989), point, and cost ratio. According to the categorical
and several other authors have evaluated the relative method, the buyer assigns either a preferred, unsatisfac-
importance of quality, cost, delivery performance, and tory, or neutral rating for each of the selected attributes
other supplier attributes. According to a review of 74 for every contending supplier. The main drawback of
articles discussing supplier selection criteria, quality was this family of methods is that the identified attributes
perceived to be most important, followed by delivery are weighted equally. The weighted point method con-
performance and cost (Weber et al. 1991). In summary, siders attributes that are weighted by the buyer. The
most of the articles referenced above suggest that man- weight for each attribute is then multiplied by the per-
agers perceive quality to be the most important supplier formance score that is assigned. Finally, these products
attribute. The conceptual articles emphasize that man- are totaled to determine a final rating for each supplier.
agers should not select suppliers based on low cost only The method, although less subjective than the former
but should consider quality, delivery performance, and method, requires that the performance measures be in
other attributes. standard (uniform) units. The cost ratio method is based
While the supplier selection literature is rich in terms on cost analysis that considers cost ratios for product
of (a) conceptual and empirical work and (b) decision quality, delivery, and customer service. The cost ratio
support methods for purchasing managers, none of the measures the cost of each factor as a percentage of total
articles cited above has studied how managers actually purchases for the supplier. The drawback of this method
choose suppliers. The past empirical articles are based is its complexity and requirement for a developed cost
on the managers’ rating of the perceived importance of accounting system. Also, the performance measures
various supplier attributes. An actual choice of supplier (cost ratios) are artificially expressed in the same units.
involves evaluating the characteristics of the suppliers Prior research has proposed various methodologies to
based on their attributes and selecting one or more sup- overcome some of the drawbacks related to these tradi-
plier(s) that best suit the needs of the firm. It can be tional methods. More specifically, numerous models have
argued that it is extremely difficult for any one supplier been developed based on particular conceptual approaches
to excel in all dimensions of performance. For example, and techniques. In their classic 1991 update of Dickson’s
a high-quality supplier might not be the one with the analysis of vendor selection decisions (Dickson 1966),
lowest-cost components. It is also possible that the com- Weber et al. reviewed the literature surrounding vendor
ponents delivered by a particular supplier excel in a few selection criteria and identified several basic techniques
quality dimensions (reliability, features) while another or models that appeared in studies over the previous
supplier might be superior in other quality dimensions 25 years. They found that the vast majority were linear
(durability, aesthetics). Therefore, an actual choice gen- weighting models, mathematical models such as eco-
erally involves trade-off among the attribute levels of nomical order quantity (EOQ), and a few probabilistic
different suppliers (Mummalaneni et al. 1996). models. Since 1991, other techniques have been applied
to the problem: analytic hierarchy process (Nydick and
DECISIONMAKING METHODOLOGIES Hill 1992; Barbarosoglu and Yazgac 1997), multi-objective
In real-world situations, there exist peer groups of programming (Weber and Ellram 1993), total cost of
vendors, which use various resources (inputs) to generate ownership (Ellram 1995), human judgments models
various results (outputs). The overall performance of sup- (Patton 1996), statistical analysis (Mummalaneni et al.
pliers may be measured by their inputs consumed and 1996), interpretative structural modeling (Mandal and
outputs generated. It is desirable to combine various Deshmukh 1994), discrete choice analysis experiment
inputs and outputs into one measure, such as the ratio (Verma and Pullman 1998), and neural networks (Siying
of aggregated outputs to aggregated inputs. This ratio et al. 1997).
can be used to evaluate and rank the performance of
Although each of these methodologies offers advan-
vendors.
tages under particular conditions, they do not provide a
Obviously, weights that reflect the relations of various generally workable methodology for combining multiple
inputs and outputs are usually needed in order to obtain criteria into a single measure of supplier performance.
such a single measure. In general, there are two possible
In the present study, an alternative methodology is
ways to determine the weights. One way is to use expert
proposed to aid purchasing managers in identifying and
opinions or other managerial information obtained
selecting suppliers. This approach attempts to address
through the delphi, analytical hierarchical process (AHP),
the need for flexible models that are highly customized
or other methods, to specify the weights. Regression or
to meet an individual firm’s particular needs. This is
optimization techniques could also be used to find the
a multi-objective approach to vendor selection that
weights.
attempts to provide a useful decision support system for
Willis, Huston, and Pohlkamp (1993) have proposed a purchasing manager faced with multiple vendors and
an effective classification of the traditional methods for trade-offs such as price, delivery reliability, and product

64 The Journal of Supply Chain Management | Spring 2000


Vendor Selection Using Principal Component Analysis

quality. An in-depth study of an actual application of β is the estimated cost of a component rejection in
this methodology is provided. the assembly line. It is a function of the percentage of
incoming components accepted by incoming quality
PRINCIPAL COMPONENT ANALYSIS control and then detected during production.
The multivariate statistical method — principal compo-
y is the estimated cost of a component rejection by
nent analysis (PCA) — is a data reduction technique used
the end user. It is a function of the percentage of
to identify a small set of variables that account for a large
incoming components accepted by incoming quality
portion of the total variance in the original variables. This
control and the production line and then returned
technique is also used to identify “latent” dimensions in
by the end user.
the data. In fact, PCA computes linear combinations of
D is the dollar value of the purchased components
variables. The first linear combination of variables accounts
in the time period.
for the largest amount of variation in the sample; the
second for the next largest amount of variance in a dimen- Compliance with the agreed date can be computed as
sion independent of the first; and so on (Bolch and Huang the moving average of delayed deliveries over the last n
1974). PCA is also a popular ranking method in multidi- procurements.
mensional analysis, but no practical application to the The operationalization of the supplier input attributes
supplier selection problem has been reported in the litera- can be carried out based on some of the selection criteria
ture. The PCA methodology has the advantage to be that are put forward in the literature. The supplier’s man-
fairly simple to exploit, since it has been available in “off- agement capabilities, production facilities and capacity,
the-shelf” computer packages for decades. and technological capability have been chosen for this
An alternative to PCA is confirmatory factor analysis application.
(CFA). Although based on a different mathematical “Management capabilities” relate to the quality of
model, CFA can be used on the same data and produces management and its operational competencies. These
similar results. dimensions may be measured on a seven-point Likert
In order to conduct the PCA process, new measures scale. In the specific application that follows, three pur-
must be defined for each input and each output. That chasing managers at the host organization were asked
is, individual output-to-input ratios for each vendor to rate each supplier. The “production facilities and
must be defined. The principal components for the capacity” attribute may be measured as the ratio
newly defined measures are then determined. Next, a between the average manufacturing capacity of the
single measure is obtained by weighting the principal supplier and the customer’s average consumption per
components in terms of the information from the unit of time. The supplier’s “technological capability”
eigenvalues. It can be seen that PCA gives the weights has many different dimensions and is difficult to mea-
among various output/input ratios defined by the sure. In this application, a questionnaire was developed
multiple outputs and multiple inputs of vendors. Finally, and submitted to the chief technical officer of each of
a performance ranking of the vendors according to the the target suppliers based on the suggestions provided
PCA scores can be constructed. by the buying organizations. The measurement indica-
tors used were the ratio of the technical staff to total
The analysis that is illustrated in the following has
staff and the ratio of the research and development
been carried out by using the SPSS computer package.
budget to sales. An overall score was obtained by calcu-
ATTRIBUTES FOR SUPPLIER SELECTION lating the sum of the relative ratios. Respondents were
Three outputs (performance indicators) and three also asked to report which of the technologies shown in
inputs (supplier attributes) are used for analysis in the Table I were adopted and implemented within their
following application. The traditional indicators of organization.
product price, shipment quality, and delivery compli-
ance are used for supplier performance. Product price is
AN APPLICATION TO VENDOR RATING IN
the average unit purchase lot price of a component. THE PACKAGING MACHINERY INDUSTRY
Shipment quality is measured using a conformity index, In order to test its effectiveness, the methodology pro-
which is built as follows: posed was applied to the supplier selection process of a
medium-sized manufacturer of bottling machinery and
αA + βB + yC complete packaging lines. The firm’s production pro-
Conformity Index =
D gram includes the design and manufacture of a range
of equipment to handle plastic containers, another
where:
group of equipment to handle glass containers, and a
α is the estimated cost of a component rejection in third type which can be used for both plastic and glass
incoming quality control. It is a function of the per- containers. Machines and bottling lines are specially
centage of defective incoming components detected designed and manufactured for the mineral water, soft
by incoming quality control. drink, fruit juice, wine, beer, liqueur, and pharmaceutical

The Journal of Supply Chain Management | Spring 2000 65


Vendor Selection Using Principal Component Analysis

Table I components used both for rinsers and for fillers. A grip
head is a device for tinplate cans and tins and glass jars
LIST OF TECHNOLOGIES ACTUALLY IN
with tinplate caps. The component is fairly complex
USE IN THE SUPPLIER COMPANIES
from a technological point of view, since it incorporates
Technologies permanent magnets with releasing grid, plate, stem, and
spring made of AISI 304 stainless steel with hard rubber
Automated Handling
pads or, alternatively, with side polyethylene rollers.
Bar Code System Among the main features that are required of this com-
Computer-Aided Design (CAD) ponent are design excellence and highly accurate con-
Computerized Numerical Control (CNC) struction in order to ensure trouble-free and quiet
Direct Numerical Control functioning, even when running at the highest produc-
Electronic Data Interchange (EDI) tion speeds.

Computer-Aided Manufacturing (CAM) More than 20 suppliers were available in the market-
place for this type of component. The analysis was car-
Computer-Aided Engineering (CAE)
ried out on 23 suppliers. Table II shows the suppliers’
Rapid Prototyping Technologies (RPT) attributes. These measures are the normalized values of
Communication Technologies the three performance factors and the three input attrib-
Computer Integrated Manufacturing (CIM) utes previously illustrated.
Automated Storage and Retrieval Systems (AS/RS) Six output/input ratios for the suppliers’ attributes
were defined:
d1 = product price/technological capability
industries. The company has an annual sales volume of d2 = product price/management capabilities
$43 million and employs 240 individuals (60 in technical d3 = shipment quality/management capabilities
departments). Its major production lines include: d4 = shipment quality/technological capability
• Rinsing machines (820 units sold worldwide) d5 = delivery compliance/production capacity
• Unscrambling machines (285 units sold d6 = delivery compliance/management capabilities
worldwide)
The decision maker would like to select the supplier
• Filling machines (60 units sold worldwide)
that provides the best combination of the performance
• Complete traditional bottling lines parameters. In statistical terms, these suppliers are
• Complete aseptic filling lines extreme observations that lie away from the rest of the
By its very nature, the purchasing function is a basic data. Thus, a procedure that would identify outliers is
part of business management, and the selection of com- needed. Principal component analysis is a procedure
petent suppliers has long been regarded as one of the that identifies outlying suppliers regardless of the
most important tasks to be performed. PCA was applied importance the purchasing manager attaches to each
to the evaluation of the performances of the suppliers performance parameter of the vendor. Most of the vari-
of the so-called “grip heads,” one of the most critical ability in the data set is contained in the first few linear

Table II
SUPPLIERS’ ATTRIBUTES (NORMALIZED VALUES)

Production
Management Facilities Technological Delivery
Supplier Capabilities and Capacity Capabilities Price Quality Compliance
1 0.622 0.261 0.667 0.958 0.100 0.122
2 0.500 0.333 0.571 1.000 0.200 0.200
3 0.737 0.429 0.400 0.935 0.133 0.167
4 0.683 0.286 0.444 0.983 0.182 1.000
5 0.452 0.353 0.400 0.958 0.400 0.040
6 0.509 1.000 0.800 0.975 0.167 0.032
7 1.000 0.500 0.571 0.943 0.333 0.179
8 0.778 0.667 0.571 0.983 1.000 0.093
9 0.596 0.176 0.444 0.920 0.167 0.060
10 0.528 0.545 1.000 1.000 0.222 0.049

66 The Journal of Supply Chain Management | Spring 2000


Vendor Selection Using Principal Component Analysis

combinations of variables, i.e., the principal components. Table III


In other words, PCA is employed to identify the principal
CORRELATION MATRIX
components that are respectively different linear combi-
nations of the performance variables so that the principal d1 d2 d3 d4 d5 d6
components can be multiplied by their eigenvalues to
d1 1.000 -0.192 0.253 -0.203 -0.278 -0.104
obtain a weighted measure of the variables.
d2 -0.192 1.000 -0.096 0.198 0.222 -0.071
The first step in PCA consists of testing whether the
d3 0.253 -0.096 1.000 0.704** -0.236 -0.121
variables show a sufficient level of correlation. To this
extent, both the correlation matrix (Table III) and the d4 -0.203 0.198 0.704** 1.000 -0.018 -0.032
Bartlett’s test of sphericity have been analyzed. The coef- d5 -0.278 0.222 -0.236 -0.018 1.000 0.900**
ficients are the usual Pearson correlations (one-tailed) d6 -0.104 -0.071 -0.121 -0.032 0.900** 1.000
and confirm that significant correlations can be found.
** Significant at the 0.001 level (one-tailed)
Bartlett’s test of sphericity is a useful instrument to
test the null hypothesis that the correlation matrix is an
identity matrix. If the null hypothesis cannot be rejected,
and the sample size is reasonably large, the decision
maker should reconsider the use of multivariate analysis,
since the dependent variables are not correlated. In this
case, the null hypothesis is rejected at the 0.001 level. Figure 1

A rule-of-thumb for determining the number of com- EIGENVALUE SCREE PLOT


ponents to extract is to consider the “eigenvalue greater
than one” criterion. The so-called “eigenvalue scree plot” 2.5
is often useful in graphically determining the number of
factors extracted (Figure 1). In the present analysis, three
components are a workable solution (since the residual
2.0
components have all eigenvalues less than 1).
Components 1, 2, and 3 account for approximately 83
percent of the total variance of the variables (Table IV).
1.5
Eigenvalue

The percentage of variance explained by each compo-


nent represents its relative importance.
The term “rotated component” in the heading in
Table IV is due to the fact that an orthogonal rotation is 1.0
usually carried out on the components that are initially
extracted. Usually, the initial component extraction
does not give interpretable results. One of the purposes 0.5
of rotation is to obtain components that can be named
and interpreted more clearly. In other words, rotation
makes the larger loadings of the different variables on
each component larger than before and the smaller 0.0 1
1 2 3 4 5 6
loadings smaller than before. There are a number of
Number of Component
different methods of extraction — here, the varimax
method has been chosen. The “rotated” loadings of each
variable on each component are reported in Table V.
The interpretation of Table V leads the decision maker
to conclude that component 1 loads on matters con-
cerning delivery compliance, component 2 loads on
matters concerning product quality, and component 3
loads on matters concerning price. Table IV
For each variable considered (d1, d2, d3, d4, d5, and TOTAL VARIANCE EXPLAINED BY THE COMPONENTS
d6), a coefficient wi (i =1 to 6) is obtained by multiplying
the loadings on each component by the percentage of Weights of the rotated component
variance explained by the component. For instance, w1 Component Total % of Variance % Cumulative
is obtained as follows: 1 1.913 31.885 31.885
w1 = 0.171 * 0.31885 + 0 * 0.28508 + 0.749 * 0.22949 = 0.2264 2 1.710 28.508 60.393
The coefficients w2 to w6 are obtained in the same way. 3 1.377 22.949 83.342
Each coefficient is then multiplied by the value of the

The Journal of Supply Chain Management | Spring 2000 67


Vendor Selection Using Principal Component Analysis

corresponding variable (d1 to d6) for each supplier to DISCUSSION ON THE MANAGERIAL
get a final supplier score (Table VI). Based on these IMPLICATIONS OF THE METHODOLOGY
scores, the final ranking is obtained. PROPOSED
Summing up, supplier number 4 ends up as the supplier In an era of global sourcing, a firm’s success often
that provides the best performances with respect to the hinges on the most appropriate selection of its suppliers.
three components identified. Supplier selection is sometimes very complicated, owing
to a variety of uncontrollable and unpredictable factors
that affect the decision. As with any decision process,
the industrial vendor selection decision involves two
basic but distinct tasks: evaluation and choice. The
Table V evaluation element typically consists of identifying the
MATRIX OF ROTATED COMPONENTS attributes, criteria, or factors relevant to the decision
(LOADINGS SMALLER THAN 0.1 ARE OMITTED) and then measuring or rating each vendor by considering
each of the relevant factors. When a supplier selection
Variable Component decision needs to be made, the buyer generally establishes
1 2 3 a set of evaluation criteria that can be used to compare
potential sources. Purchasing managers use all available
d6 0.987
attributes when evaluating supplier performance.
d5 0.946 0.247 Vendor selection decisions vary according to the pur-
d3 0.115 0.923 0.256 chasing situation. More specifically, Patton (1996) has
d4 0.918 0.282 shown that the predictive ability of the vendor selection
d2 0.778 models and the relative importance placed on evalua-
d1 0.171 0.749 tive criteria vary largely in accordance with the nature
of the selection situation. In other words, there may not
be a generalized consensus on how to weight the rela-
Table VI tive relevance of the different criteria since they are
FINAL RANKING OF SUPPLIERS highly firm- and situation-specific. Nevertheless, a crit-
ical part of the overall supplier selection process is the
SCORE SUPPLIER determination of the relative importance of each of the
NUMBER factors (Ellram 1990). Further, in making their choices,
2.292 4 the purchasing managers must necessarily make trade-
1.798 14 offs among different levels of these attributes.

1.658 5 The methodology presented in this article may help to


find a solution to these problems since:
1.639 8
• The relative importance of each vendor
1.622 17
attribute and performance is determined in an
1.482 11
objective manner.
1.396 18 • Unlike most mathematical methods, individual
1.391 22 judgments and measures (i.e., Likert-type ques-
1.356 2 tionnaires) may be effectively used to collect the
1.271 12 relevant data.
1.241 20 CONCLUSIONS
1.224 13 This article describes a multiple-attribute approach
1.209 9 based on the use of principal component analysis,
1.180 19 aimed at helping purchasing managers to formulate
1.168 23 viable sourcing strategies in the changing marketplace.
An application of the methodology using actual data
1.151 16
retrieved from a firm operating in the bottling industry
1.091 3 is illustrated. PCA has proved to be capable of handling
1.066 21 multiple conflicting attributes inherent in supplier selec-
1.013 7 tion while simultaneously trading-off key supplier selec-
0.974 15 tion criteria.
0.937 10 PCA’s strength is in simultaneously considering multiple
0.915 1 inputs and multiple outputs without any need for a priori
assignment of weights. PCA also has a distinct advantage
0.857 6

68 The Journal of Supply Chain Management | Spring 2000


Vendor Selection Using Principal Component Analysis

over the traditional methods of evaluating suppliers’ Hakansson, H. and B. Wootz. “Supplier Selection in an
performances in that it is not necessary to state the International Environment: An Experimental Study,” Journal
of Marketing Research, (12), 1975, pp. 46-51.
performance measures in the same units. The relevant
Jackson, G.C. “Just-In-Time Production: Implications for
attributes of suppliers can, in fact, be measured in any Logistics Managers,” Journal of Business Logistics, (4), 1983,
unit such as money, percentages, and qualitative subjec- pp. 1-19.
tive judgments. Furthermore, since PCA does not force Karmarker, U.S. “Integrative Research in Marketing and
the decision maker into a standardized rating for the Operations Management,” Journal of Marketing Research, (33),
1996, pp. 125-133.
attributes, this technique is less subjective than the
traditional methods. Kraljic, P. “Purchasing Must Become Supply Management,”
Harvard Business Review, (61), 1983, pp. 109-117.

REFERENCES Mandal, A. and S.G. Deshmukh. “Vendor Selection Using


Ansari, A. and B. Modarress. “JIT Purchasing as a Quality and Interpretative Structural Modelling (ISM),” International Journal of
Productivity Center,” International Journal of Production Operations and Production Management, (14:6), 1994, pp. 52-59.
Research, (26), 1980, pp. 19-26. Miller, J.G. “Production/Operations Management: Agenda for
Ansari, A. and B. Modarress. “Just-In-Time Purchasing: the ’80s,” Decision Sciences, (12), 1981, pp. 547-571.
Problems and Solutions,” Journal of Purchasing and Materials Monczka, R.M., L.C. Giunipero, and R.F. Reck. “Perceived
Management, (22:2), Summer 1986, pp. 11-15. Importance of Supplier Information,” Journal of Purchasing and
Banker, R.D. and I.S. Khosla. “Economics of Operations Materials Management, (17), 1981, pp. 21-29.
Management: A Research Perspective,” Journal of Operations Mummalaneni, V., K.M. Dubas, and C. Chao. “Chinese
Management, (12), 1995, pp. 423-425. Purchasing Managers’ Preferences and Trade-Offs in Supplier
Barbarosoglu, G. and T. Yazgac. “An Application of the Selection and Performance Evaluation,” Industrial Marketing
Analytic Hierarchy Process to the Supplier Selection Problem,” Management, (25:2), 1996, pp. 115-124.
Production and Inventory Management Journal, First Quarter Nydick, R.L. and R.P. Hill. “Using the Analytic Hierarchy
1997, pp. 15-21. Process to Structure the Supplier Selection Procedure,”
Benton, W.C. and L. Krajeski. “Vendor Performance and International Journal of Purchasing and Materials Management,
Alternative Manufacturing Environments,” Decision Sciences, (28:2), Spring 1992, pp. 31-36.
(21), 1990, pp. 403-415. Patton, W.W. “Use of Human Judgment Models in Industrial
Bernard, P. “Managing Vendor Performance,” Production and Buyer’s Vendor Selection Decisions,” Industrial Marketing
Inventory Management Journal, (30), 1989, pp. 1-7. Management, (25), 1996, pp. 135-149.

Bolch, B.W. and C.T. Huang. Multivariate Statistical Methods for Pearson, J.N. and L.M. Ellram. “Supplier Selection and
Business and Economics, Prentice-Hall, Englewood Cliffs, NJ, Evaluation in Small versus Large Electronics Firms,” Journal of
1974. Small Business Management, (33:4), 1995, pp. 53-65.

Browning, J.M., N.B. Zabriskie, and A.B. Huellmantel. Sheth, J.N. “A Model for Industrial Buyer Behavior,” Journal of
“Strategic Purchasing Planning,” Journal of Purchasing and Marketing, (37), 1973, 50-56.
Materials Management, (19:1), Spring 1983, pp. 19-24. Siying, W., Z. Jinlong, and L. Zhicheng. “A Supplier-Selecting
Burton, T.T. “JIT/Repetitive Sourcing Strategies: ‘Tying the System Using a Neural Network,” 1997 IEEE International
Knot’ with Your Suppliers,” Production and Inventory Conference on Intelligent Processing Systems, IEEE, New York,
Management Journal, (29), 1988, pp. 38-41. NY, pp. 468-471.

Cardozo, R.N. and J.W. Cagley. “Experimental Study of Treleven, M. “Single Sourcing: A Management Tool for the
Industrial Buyer Behavior,” Journal of Marketing Research, (8), Quality Supplier,” Journal of Purchasing and Materials
1971, pp. 329-334. Management, (23:1), Spring 1987, pp. 19-24.

Chapman, S.N. “Just-In-Time Supplier Inventory: An Empirical Verma, R. and M.E. Pullman. “An Analysis of the Supplier
Implementation Model,” International Journal of Production Selection Process,” Omega, (26:6), 1998, pp. 739-750.
Research, (27), 1993, pp. 1993-2007. Wagner, J., R. Ettenson, and J. Parrish. “Vendor Selection
Chapman, S.N. and P.L. Carter. “Supplier/Customer Inventory among Retail Buyers: An Analysis by Merchandise Division,”
Relationships under Just-In-Time,” Decision Sciences, (21), Journal of Retailing, (65), 1989, pp. 58-77.
1990, pp. 35-51. Weber, C.A., J.R. Current, and W.C. Benton. “Vendor Selection
Dempsey, W.A. “Vendor Selection and the Buying Process,” Criteria and Methods,” European Journal of Operational Research,
Industrial Marketing Management, (7), 1978, 257-267. (50), 1991, pp. 2-18.

Dickson, G. “An Analysis of Vendor Selection Systems and Weber, C.A. and L.M. Ellram. “Supplier Selection Using Multi-
Decisions,” Journal of Purchasing, (2), 1966, pp. 28-41. Objective Programming: A Decision Support System Approach,”
International Journal of Physical Distribution and Logistics
Ellram, L. “The Supplier Selection Decision in Strategic Management, (23:2), 1993, pp. 3-14.
Partnerships,” Journal of Purchasing and Materials Management,
(26:4), Fall 1990, pp. 8-15. Willis, T.H., C.R. Huston, and F. Pohlkamp. “Evaluation
Measures of Just-In-Time Supplier Performance,” Production
Ellram, L. “Total Cost of Ownership: An Analysis Approach and Inventory Management Journal, Second Quarter 1993,
for Purchasing,” International Journal of Physical Distribution pp. 1-6.
and Logistics Management, (25:8), 1995, pp. 4-23.
Wilson, E.J. “The Relative Importance of Supplier Selection
Flynn, B.B., R.G. Schroeder, and S. Sakakibara. “A Framework Criteria: A Review and Update,” International Journal of
for Quality Management Research and an Associated Purchasing and Materials Management, (30:3), Summer 1994,
Measurement Instrument,” Journal of Operations Management, pp. 35-41.
(11), 1994, pp. 339-366.
Zhu, J. “Data Envelopment Analysis vs. Principal Component
Hahn, C.K., P.A. Pinto, and D.J. Bragg. “‘Just-In-Time’ Analysis: An Illustrative Study of Economic Performance of
Production and Purchasing,” Journal of Purchasing and Chinese Cities,” European Journal of Operational Research, (111),
Materials Management, (19:3), Fall 1983, pp. 2-10. 1998, pp. 50-61.

The Journal of Supply Chain Management | Spring 2000 69

You might also like