Chapter 07 - The construction of a toolbox for supplier selection in

advertisement
VII The construction of a toolbox for
supplier selection in four steps
In this chapter we describe the process of constructing a toolbox for supplier
selection. The position of this chapter in the overall stepwise planning is depicted in
figure 7.1.
Development of a framework for
analysing decision making (III)
Functional requirements of
purchasing decision models
Functional requirements of
purchasing decision models
Analysing purchasing literature
with this framework (IV)
Evaluation of OR and System
Analysis models (VI)
Evaluation of available
purchasing decision models (V)
Designing a toolbox for
supporting supplier selection
(VII)
Empirical testing of the toolbox
(VIII + IX)
Evaluation of the toolbox (X).
Conclusions (XI)
Figure 7.1.: Positioning of chapter VII
As pointed out in chapter II, the toolbox should make it possible to link
purchasing decision situations (and after chapter IV more specifically supplier selection
situations) to adequate and useful means of decision support. The construction follows
four steps. The following sections respectively deal with these steps. In the first section
we define the starting points and the design specifications for the toolbox as well as the
tools (i.e. the prescriptive decision models) in it. The second and third sections deal with
the question how to place compartments and subcompartments in the toolbox. Finally, in
the fourth section we discuss the assignment of the available tools (which were identified
in the previous two chapters) to these (sub) compartments.
Chapter VII: The construction of a toolbox for supplier selection in four steps
First we define starting points and design specifications
The design of the toolbox that results from this chapter is based on a number
of starting points and specifications. To a large extent, these starting points and
specifications follow directly from the findings in the previous chapters. However, some
additional starting points and specifications are of a more personal nature. Together they
determine the shape of the toolbox as well as its contents.
The starting points cover basic assumptions
The complexity and importance of the supplier selection are the ultimate determinants
for choosing decision models
In chapter II, we defined a decision model as an explicit representation form
for someone’s view on a decision problem. In this thesis, a basic starting point is that
whether or not such a decision model is appropriate for supporting a particular decision
problem, and therefore also for supplier selection, basically depends on the following
two factors:
1.
The perceived importance of the decision; the basic question here is: "Does
the importance of the decision justify the investment of time, effort and other
resources required in building and using the model?"
2.
The perceived complexity of the decision; the basic question here is: "Does
the decision model effectively deal with the specific properties the decision
maker feels are complicating his or her decision?".
Although various authors on this subject suggest a variety of factors to be
taken into account (see for example Oxenfeldt, 1979; Pinsoneault and Kraemer, 1989),
importance and complexity appear to be the basic underlying determinants. Complexity
refers to the matching between the type of support decision aids offer and the cognitive
needs the decision maker(s) may have in a specific decision situation. Importance plays a
role because the decision-maker should assess whether or not the expected benefits from
using a decision aid exceed the costs of building it (Timmermans, 1991).
Furthermore, we argue that the evaluation of existing and possibly useful
decision models with respect to the two determinants discussed in the previous
subsection contains technical/objective aspects as well as subjective aspects. For
example, if the purchasing situation at hand concerns a so-called new-task situation for a
one-time service, we may a priori (objectively) argue that e.g. the mathematical
programming model of Weber & Ellram (1992) is not appropriate. The purchasing
situation concerns a service that will be rendered once whereas the mathematical
programming model deals with several items that are supplied on a regular basis.
Therefore, the mathematical model assumes that we have all kinds of historical data, e.g.
delivery performance of the supplier, which is clearly not possible in a new task
situation. We argue that to some degree we can objectively assess the incompatibility of
a (group of) decision model(s) with a purchasing situation, through a priori (theoretical
reasoning) and/or through empirical experiments. The final toolbox will be based on
these theoretical reasonings and empirical experiments. The results of the theoretical
158
Chapter VII: The construction of a toolbox for supplier selection in four steps
reasonings are presented later on in this chapter. The results of the empirical experiments
are discussed in the next chapter.
However, the final choice for a particular decision model is made on the basis
of subjective complexity and importance considerations, which we consider to be
determined by an individual’s cognitive skills, belief and ‘view of the world’. These
subjective aspects are therefore not incorporated in the final toolbox.
We assume that some structure is perceived in the initial supplier selection situation
Following the general starting points as described in chapter II, we argue that
supplier selection decisions may vary from structured decisions to semi-structured
decisions (in terms of Simon’s framework). Once the decision makers think of a
situation or a problem as a supplier selection problem, or are explicitly urged to do so,
the problem to some extent seems rather structured, at least in the sense that, irrespective
of possible underlying motives, the type of possible courses of action has been
determined, i.e. selecting one or more suppliers, now or at some point in the future. In
designing our toolbox we assume that this (mental) awareness exists. The use of decision
models may from this point on be aimed at a) re-examining the justification of this
awareness, i.e. checking whether or not the ultimate problem is really solved by
choosing one or more suppliers and/or b) supporting the process of formulating criteria
for selecting suppliers as well as c) the actual selection itself.
The toolbox should not per se lead to the prescription of one best decision model
In line with our discussion on objective and subjective aspects, it would seem
unrealistic and perhaps even improper to aim at constructing a toolbox that almost by
definition and ultimately leads to the recommendation of one 'best' model for a specific
situation. As many authors, among which Oxenfeldt (1979) and Bana e Costa et al
(1997), point out that especially in the case of important decisions, it is wise to employ
several different approaches to the same decision. Therefore, we will not strive at a
toolbox which ‘forces’ the user to ultimately choose one and only one decision model.
Instead, the toolbox will often suggest a set of tools to the purchaser, from which a final
choice can be made. The decision models finally offered to the purchaser may differ in
terms of technicalities but not in terms of function. This may very well imply the use of
more than one decision model for one particular supplier selection situation.
The design specifications indicate the desired functions and characteristics
of the toolbox
The toolbox should accommodate the diversity present in supplier selection situations
Already from the statement of this thesis’s objectives (see chapter II) it
becomes clear that the toolbox should lead the purchaser to appropriate prescriptive
decision models given a specific supplier selection situation. More specifically, chapter
IV has led us to conclude that not only there are several levels of supplier selection (see
figure 5.3) but also that prescriptive decision models for supplier selection should enable
the modelling of different degrees (and forms) of complexity. Furthermore we also
concluded that the various decision models in the toolbox should cover the variance in
the importance of supplier selection decisions. Clearly, the toolbox as well as the
159
Chapter VII: The construction of a toolbox for supplier selection in four steps
decision models in the toolbox should accommodate these three dimensions of variety in
supplier selection when guiding the purchaser in finding useful decision models.
The toolbox should cover all decision-making phases in supplier selection
In chapter III we found that a prescriptive approach towards decision making
best serves our research objectives. As we already described in detail in chapter III and
IV, such a prescriptive approach also considers support for the phases that precede the
actual choice from among some alternative suppliers. These phases involve what we call
‘problem definition’ and this includes such activities as defining search boundaries,
objectives, criteria and the alternatives to choose from. In our analysis in chapter IV of
supplier selection literature, we concluded that the decision models (and therefore also
the toolbox) should enable us to place appropriate emphasis on these different phases in
supplier selection, e.g. finding out exactly what we want to achieve by selecting a
supplier, what the criteria for evaluation of suppliers should be and how this evaluation
should take place. The toolbox should explicitly recognise the problem definition phases
in addition to the ultimate choice phase in supplier selection.
The toolbox should be sufficiently comprehensible and manageable for practical use
The analysis in chapter V and VI resulted in a great number of decision models.
In addition, in the previous subsections we addressed the high degree of diversity in
supplier selection situations. In the design of the toolbox we will have to find a balance
between on the one hand recognising this spectrum of situations and at the other hand
keeping the toolbox sufficiently comprehensive, accessible and manageable for possible
practical use. This implies that we will have to take into account that the number of
supplier selection situations distinguished in the toolbox should be high enough to
represent the existing diversity in supplier selection while at the same time the number of
supplier selection situations posed to the purchaser should be low enough to assure easy
comprehension and overview.
The toolbox should follow existing purchasing theories
As a general principle, we believe we should preferably try to follow as much
as possible the existing purchasing theory as far as the modelling of diversity in supplier
selection is concerned. This means that when designing the toolbox we will first try to
use existing classifications of purchasing situations and if necessary, complement the
toolbox with new insights.
We construct compartments in the toolbox to represent
different situations
The first design specification addressed in the previous section concerned
accommodating the diversity present in supplier selection. We do this by creating useful
compartments in the toolbox. In the final phase of the design, these compartments are
filled with appropriate decision models. Each compartment then represents a particular
supplier selection situation. More specifically, the preceding chapters (III) and (IV) have
made clear that three dimensions of differences between supplier selection situations can
160
Chapter VII: The construction of a toolbox for supplier selection in four steps
be discerned: (1) differences regarding the (trigger-) type of supplier selection, (2)
differences regarding the complexity of the supplier selection situation and (3)
differences regarding the importance of the supplier selection. The compartments are
constructed along these three dimensions. The dimensions are discussed in the following
sections.
The compartments cover different types of supplier selection
In chapter IV we distinguished between several levels and areas of purchasing
decision making. From that perspective, supplier selection can be seen to take place on
two levels: (1) on what we called the initial purchasing level and (2) on the so-called
adaptive control level. The first step in constructing compartments in the (still empty)
toolbox consists of separating such initial supplier selections from adaptive supplier
selections.
Adaptive
supplier
selections
Initial
supplier
selections
Figure 7.2: Division into initial and adaptive supplier selections
161
Chapter VII: The construction of a toolbox for supplier selection in four steps
Initial supplier selections are reactive and triggered by internal customers
The essential characteristic of initial supplier selections is that they are
initiated by an internal customer’s request for a supplier for a particular item or service.
In that respect we call initial supplier selections reactive. The specification of the item or
service, which may be new or modified, drives the search and final selection of a
supplier.
Specify
Order
Customers,
suppliers,
management
Initial supplier
selection
Select supplier
Close
contract
Figure 7.3: Illustration of initial supplier selection
In case of perceived new future requirements, purchasing may be triggered to
identify and pre-select suppliers while there may not yet be a very detailed specification.
Adaptive supplier selections are proactive or case-independent and triggered otherwise
The essential characteristic of adaptive supplier selection decisions is that
drivers other than a new or modified (specific) need from an internal customer, trigger
these decisions. In that respect we call adaptive supplier selections proactive. In
addition, adaptive supplier selections always relate to existing or known suppliers. We
also call adaptive supplier selections case-independent. This means that the selection as
such is not directly related to a specific case concerning a request from an internal
customer. Examples of such (proactive and case-independent) drivers are the following
(Van Stekelenborg and De Boer, 1995):
162
Chapter VII: The construction of a toolbox for supplier selection in four steps
1.
2.
3.
4.
Replacing an existing supplier because of its unsatisfactory supply
performance (while there is no need to change or modify the
specification);
Reducing a large group of (often called ‘approved’) suppliers to a
smaller number in order to improve purchasing efficiency;
Certain market developments, e.g. emerging regions, entry of new
suppliers or decreasing prices may trigger adaptive supplier selections;
Time may also be such a trigger, e.g. every two years the supply
market may be scanned to evaluate the appropriateness of the existing
set of suppliers.
Specify
Order
Customers,
suppliers,
management
Adaptive supplier
selection
Select supplier
Close
contract
Figure 7.4: Illustration of adaptive supplier selection
The compartments cover different levels of complexity and importance
So far we have created separate compartments for initial and adaptive supplier
selections. In this section and in accordance with the first design specification, we
discuss the creation of additional compartments in order to distinguish between supplier
selections that differ in terms of complexity and importance. From chapter IV we recall
that complexity relates to such factors as: the number of suppliers involved, the level of
uncertainty, the number and nature of the selection criteria etceteras.
We use the Buy-grid model to cover different levels of complexity
Several authors, mostly from the field of Industrial Marketing, have studied
complexity in purchasing and supplier selection (Fisher, 1970; Faris et al., 1967; and
Bunn, 1993).
Robinson and Faris argue that purchasing situations can be classified into
three different levels of complexity. The three levels are: new task situation, modified
163
Chapter VII: The construction of a toolbox for supplier selection in four steps
rebuy situation and straight rebuy situation. Some of the typical characteristics of these
situations are given in the table below.
- entirely new product/service; no previous
experience
- no (known) suppliers
- high level of uncertainty with respect to
the specification
- extensive problem solving; group decision
making
- new product/service to be purchased from
Modified rebuy
known suppliers
- existing(modified) products to be
purchased from new suppliers
- moderate level of uncertainty with respect
to specification
- less extensive problem solving
- perfect information concerning
Straight rebuy
specification and supplier
- involves placing an order within existing
contracts and agreements
Table 7.1: Classification of purchasing situations (Faris et al., 1967)
New task situation
Obviously, new task situations are the most complex, at least in the sense that
the level of uncertainty is the highest compared to the modified rebuy and the straight
rebuy. We now investigate how we can incorporate the distinction between new task,
modified rebuy and straight rebuy in the toolbox. First, we relate this classification to the
current compartments.
Initial supplier
selections
X
X
Adaptive
selections
supplier
New Task
Modified rebuy
X
Straight rebuy
X
Table 7.2: supplier selection related to three typical purchasing situations
New task purchases by definition involve initial supplier selections because
clearly the need (i.e. the specification) is completely new and this specification is the
direct and immediate trigger for the supplier selection process.
Modified rebuy situations may relate to initial as well as adaptive supplier
selections. If the modified rebuy concerns the purchase of a new product from existing
suppliers, this obviously is an initial selection as the (new) specification drives the
selection. However, if the modified rebuy involves buying a similar (known) product
from a new supplier, this is not the case and we are consequently dealing with an
adaptive supplier selection.
Finally, straight rebuy situations are related to adaptive supplier selections.
Strictly speaking, the straight rebuy situation does not include supplier selection.
Therefore, we speak of ‘relating to’ instead of ‘including’ or ‘covering’. More
specifically, ‘relating to’ must be interpreted as ‘following’ or ‘preceding’. Straight
rebuys basically consist of placing orders. The supplier to use is not an issue. Compared
to a modified rebuy situation, the specification is even more stable and known, which
164
Chapter VII: The construction of a toolbox for supplier selection in four steps
implies that there are no (specification-)changes, which could trigger a supplier
selection. From time to time, however, the existing supplier(s) may be evaluated and
possibly replaced. The latter decision as such however, is not part of the straight rebuy
(strictly speaking it is a modified rebuy) yet it follows a series of straight rebuys and
marks the beginning of a new series. Supplier selections that possibly follow from this
are thus adaptive supplier selections.
Adaptive
supplier
selections
Initial
supplier
selections
Straight
rebuy
New
task
Modified
rebuy
Figure 7.3: Incorporation of new task, modified rebuy and straight rebuy situations
The incorporation of the distinction between new task, modified rebuy and
straight rebuy into the toolbox is depicted in figure 7.3.
We use Kraljic’s portfolio model to cover different levels of risk and importance
The distinction between new task, modified rebuy and straight rebuy
facilitates a recognisable ‘entrance’ for the purchaser and at the same time the
classification comprises different levels of uncertainty about the purchase and the
accompanying supplier selection. However, the classification does not cover other
dimensions of complexity, e.g. the number of suppliers and criteria involved in the
selection. In addition, the other determinative factor when choosing decision models, i.e.
the importance of the supplier selection, has not been recognised at all so far.
A contribution that may prove useful in covering additional dimensions of
complexity as well as importance is Kraljic’s (1983) portfolio approach. In this portfolio,
the perceived importance and complexity of a purchasing situation is identified in terms
of two factors: profit impact and supply risk. Profit impact includes such elements as:
165
Chapter VII: The construction of a toolbox for supplier selection in four steps
-
(expected) monetary volume involved with the goods and/or services to be
purchased;
(expected) percentage of the total purchase cost or budget;
impact on (future) product quality;
impact on business growth.
Indicators of supply risk may include:
-
availability of the goods/services under consideration;
number of potential suppliers;
competitive demand;
make-or-buy opportunities;
substitution possibilities.
Depending on the scores on these factors, purchases (and therefore the related
supplier selection decisions) can be grouped according to Kraljic's classification into
strategic, bottleneck, leverage and routine purchases. Court and Steele (19-) propose a
similar classification, which distinguishes between ‘strategic critical’, ‘strategic
security’, ‘tactical profit’ and ‘tactical acquisition’, respectively. This is illustrated in
table 7.3.
Low profit
impact
High profit
impact
Low supply risk
Routine items;
. many suppliers
. rationalize purchasing procedures
. systems contracting
. automate/delegate
High supply risk
Bottleneck items
. monopolistic supply market
. long term contracts
. develop alternatives (internally)
. contingency planning
Leverage items
. many suppliers available
. competitive bidding
. short term contracts
. active sourcing
Strategic items
. few (difficult to switch) suppliers
. medium/long term contracts
. supplier development/partnership
(develop alternatives ‘externally’)
. continous review
Table 7.3: Purchasing portfolio matrix (based on Kraljic, 1983 and Court and Steele, 19-)
In table 7.3 some key-words of the appropriate purchasing strategies are
given. Incorporating the portfolio model into the toolbox implies that we more
specifically model the complexity of the supplier selection (at least in terms of the
number of suppliers to choose from) as well as the importance of the selection. Let us
now discuss how the portfolio classification relates to the current compartments in the
toolbox.
However, first it should be noted, that Kraljic’s model (implicitly) only
considers repetitive purchases, or more strictly put: items that are physically supplied in
a more or less repetitive fashion. This means that the new-task compartment does relate
to neither routine, bottleneck, leverage nor strategic items. The new-task compartment
covers one-off purchases (e.g. investments, projects) as well as first-time purchases of
items (or services) that in future might be purchased again. It also means that supplier
selection for routine, bottleneck, leverage and strategic items concerns modified rebuys
166
Chapter VII: The construction of a toolbox for supplier selection in four steps
and/or straight rebuy situations. We will argue that the purchasing of routine items,
bottleneck items and strategic items involves a straight rebuy situation rather than a
modified rebuy situation, albeit for different reasons.
In case of a routine item, there are many suppliers that could supply the item.
However, because of the low value of the item, it does not pay off to frequently search
for and select suppliers. Moreover, usually a whole set of related routine items (e.g.
stationary items) is assigned to one (or two) suppliers in order to achieve a highly
efficient ordering and administration procedure. The choice of the supplier is fixed for a
reasonable period of time. Intermediate changes in the desired or required items are dealt
with by the current supplier. Irrespective of such specific changes in the items requested
and/or actually purchased, the appropriateness of the supplier is reconsidered
periodically and if necessary a new (adaptive) selection will take place.
In case of bottleneck and strategic items, the choice of the supplier is also
more or less fixed. Small changes in the specification of the items are automatically dealt
with by the existing supplier. However, the reason for this is very different from the
reason in the routine-case. This time, there are virtually no suppliers to (immediately)
choose from, either because of a highly unique specification (i.e. a very strong resource
tie between the buying company and the supplier) or because of the scarcity of the
material.
Furthermore, we hold forth that leverage items typically involve modified
rebuy situations. The choice of the supplier is hardly fixed. There are many suppliers to
choose from while the high value (and saving potential) of the items justifies proactive
search and frequent selection of suppliers.
Adaptive
supplier
selections
Initial
supplier
selections
Routine items
Strategic/bottleneck
items
Leverage items
New task
Modified
rebuy
Straight
rebuy
Figure 7.4: Incorporation of Kraljic’s portfolio model into the toolbox
167
Chapter VII: The construction of a toolbox for supplier selection in four steps
The incorporation of Kraljic’s model in the toolbox is shown in figure 7.4. In
addition, we split the new task compartment in a high-importance compartment and a
low-importance compartment in order to cover the aspect of importance for new task
supplier selections.
We construct sub-compartments to cover the different
phases in each situation and enable planning of these
phases
At this point in the design process, several design specifications have been
addressed. By constructing various compartments in the toolbox, the diversity present in
supplier selection is taken into account. In addition, classifications from the purchasing
literature are used for this. Using these well-known classifications of purchasing
situations also contributes to keeping the toolbox comprehensible and manageable for
practical use. However, so far (and referring to the second design specification) the
various decision making phases in supplier selection, i.e. the problem definition up to the
final choice of a supplier, have not been included in the toolbox yet. We do this in the
following subsections by creating separate subcompartments for each decision making
phase.
The subcompartments cover four basic decision making phases
Generally, supplier selection is explicitely and rather extensively covered in
purchasing textbooks (see for example Leenders & Fearon, 1992; Scheuing, 1989; Baily
& Farmer, 1990; Heinritz et al., 1991; Van Weele, 1997). Although the various
contributions differ slightly with respect to the terms used and the level of depth, a
general two-stage pattern can be discerned. In this pattern a distinction is made between
(first) arriving at a set of acceptable suppliers which usually is added to a list of so-called
approved suppliers and secondly the ultimate choice from this approved vendor list
(AVL) for a particular supplier-product/service combination. The AVL serves as a kind
of 'stock' of basically acceptable suppliers. However, the final choice is made based upon
quotations provided by the approved vendors. In that sense, there is a disconnection
between (1) selecting an ultimate supplier-product/service combination and (2) preselecting acceptable suppliers. Referring to the generic model of decision making
discussed in chapter III as well as the starting points for this chapter, we emphasise that
two other important phases should be included: the phase of problem definition and the
formulation of criteria.
The first phase involves the definition of the supplier selection problem
This phase consists of ‘deciding’ or checking whether or not a certain problem
is solved (or a purpose served) by selecting one or more suppliers. It thus involves
determining what the ultimate problem or purpose is and why supplier selection is an
appropriate course of action.
168
Chapter VII: The construction of a toolbox for supplier selection in four steps
The next phase concerns the formulation of criteria
Once the supplier selection problem has been defined, criteria for the selection
of the supplier(s) are formulated.
Qualification and selection are subsequent phases
We already indicated that the purchasing literature generally distinguishes
between (a) arriving at a set of acceptable suppliers and (b) making a final choice from
these suppliers. For example, Misker (1996) identifies the following situations:
1.
Recognition: selection of suppliers with the purpose of creating or updating an
approved vendor list. From such a list, suppliers will be selected for
frequently recurring and important purchases;
2.
Pre-qualification: selection of suppliers with the purpose of creating a set of
suppliers
from which subsequently a supplier will be selected for a
specific, yet seldomly occurring and important purchase;
3.
Direct selection of suppliers without prior pre-qualification and resulting from
a specific requirement;
4.
Final selection of a supplier from an approved vendor list and resulting from a
specific requirement.
The explicit distinction between prequalification of suppliers and
subsequently a final choice for one of these suppliers corresponds with the distinction
made in the EC-directives on Public Procurement. Although Misker identifies ‘direct
selection’ (without pre-qualification) we argue that, at least theoretically speaking, there
will always be some form of qualification. Implicitly, it is decided which suppliers (from
all possible suppliers) will be asked to submit a quotation. Nevertheless, purchasers will
not perceive this implicit qualification as a qualification in the usual sense. Moreover,
one could argue that sometimes the supply market itself (e.g. in the form of supplier
catalogs, Yellow Pages, associations of accredited suppliers etceteras.) constitutes an
'approved vendor list'. In the sequel of this thesis we use the following definitions of
supplier qualification and supplier selection. Qualification involves bringing down the
set of ‘all’ suppliers down to a smaller set of acceptable suppliers 1. This process may be
carried out in more than one step. However, the first step always consists of defining and
determining the set of acceptable suppliers while possible subsequent steps serve to
reduce the number of suppliers to consider. Selection consists of making a final choice
from (or ranking of) the set of qualified suppliers 2.
1
This step is sometimes also referred to as ‘pre-qualification’. It may result in a so-called ‘bidders-list’ (in
one-off purchases) or an Approved Vendor list (for repeating purchases).
2
Quotations are often used in this step, although these may also be used in qualification.
169
Chapter VII: The construction of a toolbox for supplier selection in four steps
# suppliers
All suppliers
Acceptable suppliers
Qualification
Selection
Figure 7.4: Qualification and selection of suppliers
The concepts of qualification and selection are graphically illustrated in figure
7.4. In each of the compartments of the toolbox, subcompartments could be made which
would represent the phases of problem definition, formulation of criteria, qualification
and selection. However, before we do this, we address the exact timing and sequencing
of these phases in the different compartments.
The subcompartments offer a planning-structure for supplier selection
Our distinction between initial supplier selections and adaptive supplier
selections already indicated the differences between supplier selections in terms of
timing. We labelled initial supplier selections as reactive and adaptive selections as
proactive. The breakdown of supplier selections into a sequence of problem definition,
formulation of criteria, qualification and selection enables us to take a closer look at the
timing and planning of these phases in each of the compartments.
A useful concept in this respect is the so-called Supplier Selection Decision
Decoupling Point (SSDDP). Based on the Customer Order Decoupling Point (Hoekstra
and Romme, 1993) we introduce this concept to show how the final division of the
toolbox into compartments and subcompartments may not only serve as a useful frame
of reference for the purchaser when assessing the supplier selection situation on hand
and looking for appropriate decision models. We show that the toolbox also offers a
structure for pro-actively categorising packages of purchased items and services and
defining and planning supplier selection processes for the various packages. In other
words: the different compartments can also be seen as decision strategies. Each strategy
is built around a different positioning of the SSDDP.
The planning structure is based on the Customer Order Decoupling Point concept
Within the field of production control and logistics management, the concept
of the Customer Order Decoupling Point (CODP) has been developed. Van Stekelenborg
(1997) first indicated the relevance of this concept for structuring purchasing (decision
170
Chapter VII: The construction of a toolbox for supplier selection in four steps
making) activities. Before we elaborate on this in detail in the following, we first briefly
discuss the original application of this concept (see e.g. Monhemius and Durlinger,
1990). In a sequence (or chain) of manufacturing, assembly and/or packaging activities,
the CODP marks the point which separates the customer-driven part of the chain from
the planned (forecasting-driven) part, see figure 7.5.
Activities based on planning and forecasting
Activities based on
actual customer orders
Customer Order
Decoupling Point
Figure 7.5: CODP-concept
The CODP marks the last (main) stock point in the chain. The customer is
directly served from this stock and no further stockpoints are placed downstream.
Upstream, i.e. left from the CODP, stock points may exist. The exact positioning of the
CODP depends upon several factors, e.g. the required delivery lead times, internal
throughput and set-up times, and the specificity of the demand. Depending on these
factors different positions of the CODP are possible, ranging from ‘engineering to order’
(i.e. the CODP is placed all the way to the left) to ‘make to stock’.
We introduce the Supplier Selection Decision Decoupling Point to mark differences in
supplier selection strategies
Van Stekelenborg (1997) already noted that, similar to sequences of
production or logistics processes, some phases in supplier selection (e.g. creating an
approved vendor list) may be carried out in advance, thereby anticipating on a future,
final decision. The results of these anticipative processes can be seen as some kind of
‘stock’. In case of an actual and specific (customer-) need or opportunity for a final
selection, the purchaser proceeds from this stock and uses its contents to arrive at a final
decision. The content of the stock depends on where this stock-point (which we will call
the Supplier Selection Decision Decoupling Point) is positioned.
171
Chapter VII: The construction of a toolbox for supplier selection in four steps
This may be at different points in the whole supplier selection process, see figure 7.6.
problem
definition
formulation
of criteria
Supplier
qualification
Supplier
selection
SSDDP
‘decide for the
customer’
SSDDP
‘select to order’
SSDDP
‘define,
formulate,
qualify and
select to
order’
Figure 7.6: Different positions of the Supplier Selection Decision Decoupling Point (SSDDP)
Different purchasing situations (as represented by means of the different
compartments in the toolbox) consequently result in different positions of the SSDDP.
For example, the more familiar a purchaser is with the situation, e.g. in case of
established products that have been purchased before, the further downstream the DDP
may be positioned. Obviously, the opposite applies for situations involving first time
purchases of new products. We can now investigate the position of the SSDDP in the
different compartments of the toolbox. By fixing the SSDDP in each compartment, such
a compartment becomes the basis for a (prescriptive) supplier selection strategy. In
addition to the various positions of the SSDDP in figure 7.6, we could identify one more
position, namely between ‘formulation of criteria’ and ‘supplier qualification’. However,
in the following subsections we argue that the three positions in figure 7.6 are most
useful and appropriate.
The SSDDP in new task strategies lies within the phase of problem definition
Due to the uncertainty about what might be purchased for the first time in the
future, it seems difficult and/or inappropriate to already formulate criteria and to qualify
and select suppliers. Therefore, the SSDDP is placed on the left-hand side of
‘formulation of criteria’. In the sequel we will refer to a New Task/One-Off (NTOO)
strategy if the SSDDP is placed there. However, the exact position does not seem
completely obvious. For certain new tasks, e.g. the replacement of existing investment
goods or the purchasing of goods that previously were made ‘in-house’, it is partly
possible to anticipate on the phase of problem definition. In such cases we can in
advance investigate and analyse under which circumstances or conditions replacement
and outsourcing respectively would be necessary or advantageous.
172
Chapter VII: The construction of a toolbox for supplier selection in four steps
Therefore, in a NTOO strategy we distinguish between two possible activities
in the phase of problem definition: (a) anticipating on a request to select a supplier for
the purchase of a certain good or service that currently is managed ‘in house’ and (b)
reacting to and investigating such a request. The SSDDP lies between these two
activities, after (a) and before (b). Both activities (a) and (b) are aimed at making sure
that the question: “Should we buy or should we not buy and do something else?” has
been raised (or will be raised) and properly answered, i.e. possible alternatives have been
considered. Selecting a supplier in order to purchase something serves a particular
purpose. The activities in the phase of problem definition consist of explicating this
purpose, identifying and exploring alternative ways of achieving it as well as recognising
related problems, activities or decisions. Graphically, the NTOO-strategy looks as shown
in figure 7.7.
SSDDP
Problem.def.
(a)
(b)
formulation
of criteria
Supplier
qualification
Supplier
selection
stock
‘decide to
stock’
‘decide from stock’
Figure 7.7: Positioning of the SSDDP in a NewTask One-Off strategy
Besides, note that a NTOO-strategy only involves initial supplier selections
(in the terminology from figure 7.3).
The SSDDP in a Modified Rebuy strategy lies between supplier qualification and
selection
In case of modified rebuys, the SSDDP can be positioned more downstream
as more is known about future supplier selections. The product or service (or at least
similar products and services) have been purchased before and there is experience as to
the suppliers of these products and services. Therefore, both the formulation of criteria
and the qualification of suppliers can be carried out separately from the final selection of
a supplier for a specific purchase. In addition, actually doing so offers several
advantages in terms of process efficiency as well as effectiveness (Brand, 1994). First,
instead of working through the whole process (of problem definition, formulation of
criteria, qualification of suppliers etceteras) each time an internal customer issues a
request for selecting a supplier, only the final selection stage needs to be carried out if
periodically (e.g. every year) the first three activities are carried out, resulting in a list of
approved suppliers. This simply saves time and effort each time a final selection has to
be made (remember that there will frequently be a need to select a supplier). Secondly,
the number of suppliers can be managed and limited which also means that the related
administrative costs can be managed. Thirdly, having several qualified suppliers ‘in
stock’, means a shorter time-to-selection once an internal customer issues a request for
this.
173
Chapter VII: The construction of a toolbox for supplier selection in four steps
Graphically, the SSDDP for the Modified Rebuy compartment looks as shown
in figure 7.8.
SSDDP
Problem.def.
(a)
(b)
formulation
of criteria
‘decide to
stock’
Supplier
qualification
Supplier
selection
stock
‘decide from stock’
Figure 7.8: Positioning of the SSDDP in a Modified Rebuy (MR) strategy
Thus, unlike the NTOO-strategy, supplier qualification and supplier selection
are de-coupled. Here, supplier qualification is what we have previously labelled an
adaptive decision while here supplier selection is what we called an initial selection (see
also figure 7.3). Furthermore, the nature of the problem definition phase in the MRstrategy is different from the problem definition phase in the NTOO-strategy.
Actually, in the MR-strategy, we should distinguish between problem
definition for the qualification phase and problem definition for the selection phase. In
figure 7.7 only the former is shown. Problem definition for the qualification phase
involves checking (or anticipating on) the expressed need to either acquire more
qualified suppliers, or to reduce the set of qualified suppliers or to replace some of the
qualified suppliers with new ones. Note that the question whether or not to ultimately
buy a (modified) product or service is not considered here (this question concerns the
problem definition of the selection phase). As the MR-strategy deals with repeating
(similar) purchases the problem whether or not to buy a (slightly) modified product is
relatively structured and straightforward. Consequently, this problem definition phase is
not explicitly covered in the toolbox.
The SSDDP in a SR-strategy lies after the supplier selection phase
The third basic strategy follows from the straight-rebuy compartment in the
toolbox. In such a (SR) strategy the SSDDP is placed all the way to the right, i.e. after
the final selection of a supplier. This thus applies to both routine items and bottleneck
and strategic items, although for different reasons. Therefore, in the sequel we will
distinguish between the SR-strategy for routine items and the SR-strategy for bottleneck
and strategic items. Following the reasoning behind the MR-strategy, the SSDDP could
be placed at least between supplier qualification and supplier selection. After all, similar
to the modified rebuy-compartment, we are dealing with repeating purchases of items
and services that are very similar. However, as we indicated earlier on, in case of routine
items, it does not pay-off to work through the whole process (of problem definition,
qualification etceteras) each time an internal customer requests the delivery of an item or
the execution of a service. Therefore, in the SR-strategy, the SSDDP is placed after the
phase of supplier selection. Again referring to the Customer-Order-Decoupling-Point
concept in the literature on Logistics and Production Control (see e.g. Hoekstra and
Romme, 1994) we might say that in the SR-strategy we ‘take over’ the customer. Unlike
the NTOO and the MR-strategy, the (internal) customer cannot directly trigger a (final)
174
Chapter VII: The construction of a toolbox for supplier selection in four steps
selection of a supplier. In that respect, he is not considered a customer anymore. Supplier
selections are made by the purchasing professionals and are made irrespective of
specific, individual customer requirements. Similar to the MR-strategy, problem
definition in the SR-strategy for routine items involves investigating the problem
whether or not to replace the current supplier rather than the problem whether or not to
buy a particular item.
In case of bottleneck and strategic items, the SSDDP is also placed after the
phase of supplier selection, however for reasons other than the reasons just discussed for
the routine items. For bottleneck and strategic items, the question to which extent we can
speak of supplier selection becomes relevant, as there are usually very few or just one
supplier available. So, even if internal customers require modified versions of products
or services, virtually no other suppliers could be chosen. In that respect, we should speak
of supplier (or supply-mode) evaluation rather than selection. Consequently, the phase of
problem definition is of a different nature here. The main issue is now how to deal with
the high level of dependency on the strategic/bottleneck supplier. If it is not possible to
find other suppliers, it becomes a matter of changing (the relationship to) the existing
supplier and/or finding other supply options or internal solutions. Problem definition
now involves checking/investigating ideas about how the supplier and/or the relationship
to the supplier should be changed. Graphically, the SR-strategy looks as is shown in
figure 7.9.
SSDDP
Problem.
definition.
formulation
of criteria
Supplier
qualification
Supplier
selection
stock
‘decide to
stock’
‘decide
from
stock’
Figure 7.9: Positioning of the SSDDP in a Straight Rebuy (SR) strategy
Besides, note that a SR-strategy only involves adaptive supplier selection
decisions3 (see again also figure 7.3).
We fill the sub-compartments with appropriate decision
models
We have now arrived at the point in the design process where most of the
design specifications have been addressed. The diversity present in supplier selection has
been taken into account by constructing different compartments in the toolbox. Existing,
established classifications from purchasing theory were used for this, thus also
contributing to the manageability of the toolbox. In the previous section we addressed
the inclusion in each compartment of the various decision-making phases by creating a
3
We once again stress that our concept of a SR-strategy acknowledges the fact that a straight rebuy as such
does not involve supplier selection. Strictly speaking, in a SR-strategy, the selection itself is part of a modified
rebuy (i.e. selecting a new supplier for existing items). The SR-strategy covers the selection (fixing) of a
supplier between series of straight rebuys.
175
Chapter VII: The construction of a toolbox for supplier selection in four steps
subcompartment for each phase. In addition, by introducing the SSDDP-concept, we
showed how each compartment can be used to structure and plan the phases in the
supplier selection process. In that respect, the compartments can be seen as decision
strategies. In the final step, we now discuss the assignment of the available decision
models (tools) to each subcompartment in the toolbox. This assignment process proceeds
along three criteria.
A first, obvious criterion for assigning the decision models to the
subcompartments is to consider the nature of the activity that is to be supported by the
tools in each particular subcompartment, i.e. defining the problem, formulating criteria,
qualifying or selecting suppliers. In addition, for each subcompartment we can
investigate to what extent the specific, technical properties of the activity in that
subcompartment (e.g. the number of suppliers to evaluate) ‘rule out’ one or more
decision models.
Secondly, the available information in each subcompartment (or the lack of
certain information, e.g. historical performance records of suppliers) may define specific
requirements as to the decision models that can be used to support the activity in that
subcompartment.
Thirdly, the expected effort required to use a certain decision model can be
evaluated (albeit it only roughly) in the light of the (relative) importance of the activity
in each subcompartment. The assignment of the decision models described in chapter V
and chapter VI follows in the remaining subsections of this chapter.
The first assignment operation involves decision models for problem
definition
In each compartment (strategy), the first subcompartment requires tools for
supporting the phase of problem definition. Already in chapter V and chapter VI (in
which we extensively discussed both existing decision models and potentially useful
decision models), we distinguished between decision models for problem definition and
decision models for the choice phase in supplier selection. Clearly, only the former
decision models should be assigned to the first subcompartments in each strategy.
In terms of technical properties of the problem definition activities, it is
difficult to identify differences between the compartments a priori. By definition, this
phase is unstructured and it does not logically follow that one compartment has
properties that are fundamentally different from the properties of the activities in the
other compartments. In the discussion of the supplier selection strategies in the previous
section, we concluded that the basic question that is dealt with in each strategy differs
(buy or not buy versus changing or not changing the supply structure). However, this
aspect as such does not necessarily imply that certain decision models for problem
definition become less appropriate in a particular strategy.
Nevertheless, when it comes to the required effort in using the decision
models, it does indeed seem appropriate to differentiate between the strategies. Within
the NTOO-strategy, we can distinguish between purchases of higher and purchases of
lower importance. In case of (relatively) low important, one-off purchase, it seems
unlikely that the time and effort invested in using decision models for problem definition
are justifiable. In addition, as these supplier selections are really one-off decisions, the
resulting decision model could only be used for this occasion. At first glance, a similar
reasoning could be followed for routine items (in a SR-strategy) as these items are also
considered to be of (relative) low importance. However, contrary to the one-off purchase
176
Chapter VII: The construction of a toolbox for supplier selection in four steps
of a single, unimportant item, the SR-strategy usually involves the repetitive purchasing
of many low-value items from one supplier. In addition, in a SR-strategy, a decision
model for problem definition can be used again (because of the repetitive nature of the
purchases and the periodical review of the supplier selection) while in NTOO-strategy
the decision models are only used once. In respect of these points, more effort in
supporting problem definition would be justified than in case of a NTOO-strategy. At the
other hand, due to the high level of familiarity and the low level of complexity of the
items involved, the need for an extensive problem definition (and more specifically
decision models for this) may be limited.
As for now, given the foregoing discussion, we assign all decision models for
problem definition to the different strategies, with the exception being the case of
unique, one-off and low-value purchases in the NTOO-strategy.
Problem
definition (1st
subcompartment)
Typical focus and
properties
NTOO –
strategy
MR-strategy
-
-
-
-
Appropriate
decision models
Use a
supplier or
not?
Varying
importance
One-off
decision
High
importance:
- VFT,
Influence
Diagrams, Cog.
Mapping,
AIDA, SFT,
WWS, FFA
SR-strategy
- routine -
SR-strategy
- bottleneck /
strategic -
Use more, Replacing
How to deal with
fewer or
the current
the supplier?
other
supplier?
High importance
suppliers?
Low/mode
Repeating
Moderate/
rate
evaluation
high
importance
imporRepeating
tance
decision
Repeating
decision
- Value Focused Thinking, Influence Diagrams, Cognitive
Mapping, AIDA, Strategy Formulation Table, WWS-analysis,
Framework for Formulation of Alternatives
Low
importance:
- WWS, FFA
Table 7.4: Allocation of decision models for problem definition to supplier selection strategies
The result of assigning the decision models for problem formulation to the
strategies for supplier selection is given in table 7.4
The second assignment operation involves decision models for the
formulation of criteria
In each compartment (strategy), the second subcompartment requires tools for
supporting the phase of formulating criteria. Already in chapter VI (in which we
extensively discussed potentially useful decision models), we specifically identified
decision models for this purpose. More specifically, these models were: Brainstorming,
Rough Sets and Interpretive Structural Modelling (ISM). Clearly, at most only these
decision models should be assigned to the second subcompartment in each strategy.
177
Chapter VII: The construction of a toolbox for supplier selection in four steps
However, depending on the characteristics of each subcompartment, a differentiation
may be required.
As to the technical properties of each subcompartment, potential differences
may exist but they are not relevant as the decision models are insensitive to the number
of suppliers or the number (and nature) of existing criteria.
With regard to the available information in each compartment, a
differentiation must however be made. Rough Sets require historical data about
comparable supplier selections. Therefore, Rough Sets are not useful in a NTOOstrategy. Furthermore, as ‘real’ selections are hardly made in a SR-bottleneck/strategic
strategy, Rough Sets also seem less appropriate here.
Thirdly, from a cost/benefit perspective, we argue that the application of ISM
in a NTOO-strategy should be limited to the more important cases, given the intractable
nature of this model.
Formulation of
criteria (2nd
subcompartment)
Typical focus and
properties
Appropriate
decision models
NTOO – strategy
MR-strategy
SR-strategy
- routine -
SR-strategy
- bottleneck /
strategic -
-
-
-
-
No historical
data on
suppliers
available
No
previously
used criteria
available
Varying
importance
Brainstorming and
ISM (latter only
for important
cases)
-
Historical
data on
suppliers
available
Previously
used criteria
available
Brainstorming,
ISM and Rough
Sets
-
Historical
data on
suppliers
available
Previously
used criteria
available
Brainstorming,
ISM and Rough
Sets
-
Historical data
on suppliers
available, yet
very few
actual
selections
Previously
used criteria
available
Brainstorming, ISM
Table 7.5: Allocation of decision models for formulation of criteria
The result of assigning the decision models for the formulation of criteria to
the strategies for supplier selection is given in table 7.5.
The third assignment operation involves decision models for the
qualification of suppliers
Earlier in this chapter we defined qualification as the process of bringing the
set of ‘all’ suppliers down to a smaller set of acceptable suppliers. In chapter V and
chapter VI we identified many decision models for the choice phase in supplier selection
that at first glance seem appropriate for qualification. Again, depending on the specific
characteristics of each subcompartment, we investigate to what extent the allocation of
the decision models to the subcompartments requires a differentiation.
Two technical properties are relevant when assigning the decision models to
the third subcompartment of each strategy. First, the strategies differ in terms of the
number of suppliers that will normally be available for qualification. Due to the high
level of uncertainty in the NTOO-strategy about the product or service to be purchased,
the purchaser’s unfamiliarity with the supplier market the initial set of ‘all’ known
suppliers is often relatively small. The latter is also true for SR-bottleneck/strategic, yet
178
Chapter VII: The construction of a toolbox for supplier selection in four steps
for another reason: there are simply no (or only a few) other possible suppliers.
Conversely, in the MR-strategy as well as in the SR-routine strategy, the initial set of
‘all’ suppliers known to the purchaser will be relatively large. The second technical
property concerns the decision rule that is applied in the qualification stage in the various
strategies. Both in the NTOO-strategy and in the SR-strategy the qualification is not
decoupled from the selection stage (contrary to the MR-strategy). Consequently, in the
NTOO and the SR-strategies, the process of qualification is primarily a sorting process,
i.e. dividing the initial set of ‘all’ suppliers into a subset of acceptable suppliers and a
subset of unacceptable suppliers, rather than a ranking process (which results in a
complete order of suppliers from worst to best). In the MR-strategy however, supplier
qualification and supplier selection are decoupled. As a consequence, qualification is not
only sorting but may also include ranking (e.g. adding one new supplier to the existing
set of qualified suppliers).
Another relevant difference between the strategies concerns the available
information. Clearly, in a NTOO-strategy there is no (or hardly any) information on
similar, previous qualification cases, while such information is indeed available in the
other strategies.
Finally, the importance of the supplier selection is a relevant factor.
Especially in the NTOO-strategy there may be significant differences between purchases
in this respect. Therefore, the more laborious decision models do not seem specifically
appropriate in cases of low or moderate importance. In table 7.6 the results of the
foregoing are summarised.
179
Chapter VII: The construction of a toolbox for supplier selection in four steps
Qualification of
suppliers (3rd
Subcompartment)
Typical focus and
properties
Appropriate
decision models
NTOO – strategy
MR-strategy
SR-strategy
- routine -
SR-strategy
- bottleneck /
strategic -
-
-
-
-
Small initial
set of
suppliers
Sorting
rather than
ranking
No historical
records
available
Categorical
model,
Conjunctive,
disjunctive and
lexicographic
screening, fuzzy
sets, outranking
Large set of
initial
suppliers
Sorting as
well as
ranking
Historical
data
available
For sorting:
Categorical
model,
Conjunctive,
disjunctive and
lexicographic
screening, fuzzy
sets, outranking,
neural networks,
DEA, cluster
analysis
For ranking:
Linear weighting,
lexicographic,
maximin,
weighted product,
decision analysis,
linear assignment,
topsis, distance
from target,
STEM, BordaKendall, CookSieford, GroupSmart
Large set of
initial
suppliers
Sorting
rather than
ranking
Historical
data
available
Categorical
model,
Conjunctive,
disjunctive and
lexicographic
screening, fuzzy
sets, outranking,
neural networks,
DEA, cluster
analysis
Very small
set of
suppliers
Sorting
rather than
ranking
Historical
data
available
Categorical,
(fuzzy)
conjunctive
Table 7.6: Allocation of decision models for qualification of suppliers
In addition, table 7.6 indicates appropriate decision models given the typical
focus and properties of each strategy. From all decision models for the choice phase in
supplier selection identified in chapter V and chapter VI the following decision models
support the sorting of suppliers:
-
categorical model;
conjunctive screening;
disjunctive screening;
fuzzy sets screening;
outranking models;
lexicographic screening;
neural networks;
DEA;
Cluster analysis.
From table 7.6 it becomes clear that in the NTOO-strategy DEA, neural
networks and cluster analysis have been left out. The reasons for this are the following.
180
Chapter VII: The construction of a toolbox for supplier selection in four steps
Neural networks has been left out because this model requires historical data on several
comparable qualification cases, something which is obviously not the case in a NTOOstrategy. Furthermore, DEA and cluster analysis are not included because these models
are especially useful in situations where the initial set of suppliers is large, which is
again not the case in the NTOO-strategy.
Cluster analysis, DEA and neural networks are included in the MR-strategy
and the SR-routine strategy for the same reasons they were left out of the NTOOstrategy. The SR-bottleneck strategy only includes (fuzzy) conjunctive screening
because there are so few (if any) real suppliers to sort. Finally, all decision models that
support the ranking of suppliers are included in the MR-strategy. These models are not
included in the other strategies because these strategies involve sorting rather than
ranking.
The fourth assignment operation involves decision models for the selection
of suppliers
The fourth subcompartment in each (main)compartment should contain tools
for the final selection from (ranking of) a set of acceptable suppliers. Many of the
decision models for the choice phase identified in chapter V and chapter VI support the
ranking of suppliers:
-
linear weighting models;
lexicographic model;
maximin and maximax;
decision analysis;
linear assignment model;
topsis;
distance from target model;
STEM;
fuzzy sets;
weighted product model;
cost-ratio, TCO models;
mathematical programming;
outranking models.
As in the previous subsections, we investigate the extent to which the
allocation of these models to the strategies requires a further differentiation.
First, a number of technical properties differ across the various strategies.
Although the number of suppliers available for the final selection will not be high in any
of the strategies, there will still be only very few or even one in the SRbottleneck/strategic strategy. Furthermore, due to the high level of uncertainty in the
NTOO-strategy, there is likely to be extensive interaction with the suppliers and many
criteria may be considered. In the MR-strategy as well as the SR-routine strategy there is
less uncertainty and consequently the interaction and the number of criteria used will be
lower. A specific technical property of the MR-strategy concerns the relation between
the supplier selection and the decision as to how to allocate the purchase volume to the
suppliers (given that a multi-sourcing structure is preferred or required). This property is
not (or at least far less prominent) present in the other strategies. After all, the NTOO-
181
Chapter VII: The construction of a toolbox for supplier selection in four steps
strategy only deals with first-time or one-off purchases while in the SR-strategy is
characterised by single and sole sourcing situations.
In relation to the level of uncertainty, the available information about (the
performance of) the suppliers differs across the compartments. In the MR-strategy as
well as the SR-strategy there is such information while in the NTOO-strategy historical
performance (and especially cost-) information about suppliers is not or only hardly
available.
Thirdly, similar to the previous allocation operation, the importance of the
supplier selection is a relevant factor. Especially in the NTOO-strategy there may be
significant differences between purchases in this respect. Therefore, the more laborious
decision models do not seem specifically appropriate in cases of low or moderate
importance. The decision models used in the MR and the SR-strategies however, can
often to a large extent be used again, which justifies some initial investment in building
them. In table 7.7 the results of the foregoing are summarised.
182
Chapter VII: The construction of a toolbox for supplier selection in four steps
Final selection
of suppliers (4th
Subcompartment)
Typical focus
and properties
NTOO – strategy
MR-strategy
SR-strategy
- routine -
SR-strategy
- bottleneck /
strategic -
-
-
-
-
-
-
-
Small initial
set of
suppliers
Ranking
rather than
sorting
Many
criteria
Much
interaction
No historical
records
available
Varying
importance
Model used
once
-
-
-
-
Appropriate
decision models
Linear weighting,
lexicographic,
maximin, linear
assignment,
topsis, distance
from target,
STEM, fuzzy sets,
weighted product,
outranking
Small to
moderate set
of initial
suppliers
Ranking
rather than
sorting
Also: how to
allocate
volume?
Fewer
criteria
Less
interaction
Historical
data
available
Model used
again
Linear weighting,
lexicographic,
maximin, linear
assignment,
topsis, distance
from target,
STEM, fuzzy sets,
weighted product,
outranking,
costratio and TCO
models, decision
analysis,
mathematical
programming
Small to
moderate set
of initial
suppliers
Ranking
rather than
sorting
Fewer
criteria
Less
interaction
Historical
data
available
Model used
again
Single
sourcing
rather than
multiple s.
Linear weighting,
lexicographic,
maximin, linear
assignment,
topsis, distance
from target,
STEM, fuzzy sets,
weighted product,
outranking,
decision analysis
-
-
-
Very small
set of
suppliers
(often only
one)
Historical
data
available
Evaluation
rather
selection
Sole
sourcing
Linear weighting,
distance from
target, fuzzy sets,
weighted product,
decision analysis,
linear assignment,
outranking
Table 7.7: Allocation of decision models for final selection of suppliers
From table 7.7 it becomes clear that in the NTOO-strategy some of the
possible decision models for the ranking of suppliers have not been included. Decision
Analysis is not included because the outcome of this decision model only has a useful
meaning if the supplier selection (and/or the physical supply of the product) takes place
frequently. Clearly, this is not the case in the NTOO-strategy. Moreover, building a
decision analysis model usually requires substantial historical data, which is hardly the
case in the NTOO-strategy. For the same reason, cost-ratio models are not appropriate.
Mathematical Programming is excluded as this model also requires historical
information about the suppliers (cost-)performance but also because mathematical
programming models are primarily useful in a situation where the supplier selection
decision is interrelated with the allocation of the purchase volume among several
suppliers. This is not the case in a first-time/one-off purchase.
In the SR-strategies also some decision models for ranking suppliers have not
been included. In the SR-routine strategy, Mathematical Programming and Cost-ratio
models are missing. This is due to the typical single-sourcing structure that is pursued in
this strategy. In a single sourcing structure, optimising volume allocation becomes
183
Chapter VII: The construction of a toolbox for supplier selection in four steps
irrelevant. Moreover, even if a multiple sourcing structure would exist, optimising the
allocation of the (relatively low) volume among the suppliers may barely justify the
model building efforts. Cost-ratio models are not specifically useful, as only historical
cost-information on the existing (single) supplier is available.
In the SR-bottleneck/strategic strategy, a reasoning similar to the foregoing
applies. The typical sole sourcing structure makes mathematical programming less
relevant (the volume allocation problem does not exist). In addition, the (very) low
number of suppliers implies that STEM, Topsis and some linear weighting models (e.g.
AHP) become inappropriate as these models ‘require’ at least three (but preferably more)
alternatives to compare.
Finally, as the SR-bottleneck strategy at this stage is about evaluation of the
sole supplier (compared to its past performance and/or an ideal profile) rather than
selection between ‘real’ suppliers, it makes sense to use all available information
(criteria) in this evaluation. Therefore, strictly non-compensatory models like the
maximin model and the lexicographic model are excluded.
Summary
In this chapter we dealt with the construction of the toolbox for supplier
selection as we set out in the objectives of the thesis. The construction process followed
four steps.
In the first step, starting points and design specifications were defined. The
starting points covered basic assumptions about the factors that determine the choice of a
decision model in a certain supplier selection situation and the characteristics of the
situations in which the box may be used. The design specifications indicate the desired
functions and characteristics of the toolbox.
In the second step, we created compartments in the toolbox to represent
different types of supplier selection situations as well as different levels of complexity
and importance.
Subsequently, within each compartment, we constructed subcompartments to
cover the different phases in supplier selection: problem definition, formulation of
criteria, qualification of suppliers and the final selection of suppliers. Also, we showed
how the subcompartments can be used to plan supplier selection processes.
Finally, we filled the subcompartments in the toolbox with the decision
models we mapped and investigated in the chapters V and VI.
184
Download