This article appeared in a journal published by Elsevier.

The attached
copy is furnished to the author for internal non-commercial research
and education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elsevier’s archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/copyright
 Author's personal copy

Omega 40 (2012) 722–737

Contents lists available at SciVerse ScienceDirect

Omega
journal homepage: www.elsevier.com/locate/omega

Spare parts classification and demand forecasting for stock control:

Investigating the gap between research and practice
Andrea Bacchetti n, Nicola Saccani
Supply Chain and Service Management Research Centre—Department of Industrial and Mechanical Engineering, Universita di Brescia, Via Branze 38, 25123 Brescia, Italy

a r t i c l e i n f o a b s t r a c t

Available online 12 October 2011 This paper investigates the gap between research and practice in spare parts management, with specific
Keywords: reference to durable goods addressed to private or professional customers. The paper provides a critical
Spare parts literature review of theoretical contributions about spare parts classification and demand forecasting
Classification for stock control. The discussion of ten case studies, then, allows to analyze the reasons for this gap,
Demand forecasting by addressing the limitations of models developed in literature, the role of contextual factors and the
Inventory management maturity in companies’ spare parts management practices. Four main directions for research are
Research–practice gap proposed in order to bridge the gap, namely: to develop integrated approaches to spare parts
management; to define contingency-based managerial guidelines, to favor the knowledge accumula-
tion process in companies, and to supplement theoretical models with practical relevance.
& 2011 Elsevier Ltd. All rights reserved.

1. Introduction Group [12], moreover, managers lament the lack a system

Service parts for products like household appliances, automo-
biles and copy machines has grown into a business worth more
than a $200 billion worldwide [1]. Spare parts inventories need
to be available at appropriate points within the supply chain, to
provide after-sales services and to guarantee the desired service
level [2]. However, several aspects concur in making demand and
inventory management for spare parts a complex matter: the high
number of parts managed [3]; the presence of intermittent or
lumpy demand patterns [4]; the high responsiveness required
due to downtime cost for by customers [5]; and the risk of stock
obsolescence [3].
Research on spare parts demand and inventory management
developed specific methods in the last decades, such as the
application of Croston’s method [6] and its successive modifica-
tions [7] to spare parts demand forecasting. On the other hand, in
recent years, companies increased investments in spare parts
management: Aberdeen Research estimated the total market size
for service parts management software at over $100 millions in
2005 [8]. Despite that, after-sales service and spare parts manage-
ment in several companies manufacturing and selling durable
consumer products has been traditionally overlooked [9], and the
management techniques adopted are often not differentiated
from the ones used for finished products or components used in
production [10]. As pointed out by Boone et al. [11] and Aberdeen
n
Corresponding author. Tel.: þ39 030 6595213.
E-mail address: andrea.bacchetti@ing.unibs.it (A. Bacchetti).
perspective in spare parts management, the weakness of supply
chain relationships, the inaccuracy of demand forecasts, and the
difficulty in maintaining an effective level of parts inventory
turns. In addition, the frequency of product innovation requires
to manage demand and inventory also for parts for which
historical demand or failure data are not available [11].
Such problems encountered in business practice suggest the
existence of a research–practice gap [3,16]. Although the issue of
a research–practice gap has been discussed in production, inven-
tory and demand management (e.g. [13–15]), very scarce empiri-
cal research has addressed the field of spare parts management.
The objective of this paper is, therefore, to investigate the
existence and the reasons for such a gap between research and
practice in spare parts management. Firstly, the paper aims
to provide a critical literature review of research, focusing in
particular on spare parts classification, demand forecasting, and
their integration with spare parts inventory control. Secondly,
through the findings from ten case studies of companies manu-
facturing durable goods addressed to private or professional
customers, we aim at analyzing the reasons for this gap. Third,
the conceptual tools used for the analysis (the contingency
perspective and a knowledge maturity model) and the discussion
in the light of the literature, allow to identify four directions for
future research in order to bridge the aforementioned gap.
The remaining of the paper is organized as follows. Section 2
provides a literature review of studies addressing spare parts
classification, demand forecasting, their integration with inven-
tory management and the gap between research and practice in
spare parts management. Section 3 describes the methodology

0305-0483/$ - see front matter & 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.omega.2011.06.008
 Author's personal copy

A. Bacchetti, N. Saccani / Omega 40 (2012) 722–737 723

adopted for the empirical research. Section 4 presents the findings Although several criteria have been proposed, very little
from ten case studies, comparing the spare parts management attention has been dedicated to identifying in which context a
practices of the analyzed companies. In Section 5, the findings are criterion is preferable to others. Only Boylan and Syntetos [10]
discussed in the light of extant literature, Section 6 identifies suggests that part criticality is more appropriate for technical
directions to reconcile research and practice, while the final systems rather than products used by private customers.
section points out some conclusive remarks and the limitations Table 1 also categorizes the literature according to the classi-
of this study. fication technique proposed (i.e. in which way parts are classified
once the criteria have been defined). Most papers adopt quantita-
tive classification techniques. The most common is the ABC
approach, proposed in 10 contributions. It is used for a single
2. Background
criterion, generally demand volume [25,27], and for multi-criteria
classifications. Different methods are proposed to implement
This section proposes a critical review of literature about spare
multi-criteria ABC classifications: matrix models [38,19], weighted
parts management. For spare parts classification (Section 2.1) and
linear optimization [21–23], artificial neural networks [20],
demand forecasting methods (Section 2.2) we undertook a compre-
weighted Euclidean distances with quadratic optimization [24],
hensive literature review in main journals dealing with operations
and the fuzzy logic [29]. A part from ABC, different quantitative
management, forecasting and operations research, identifying,
classification techniques are developed. A demand-based classifi-
respectively, 25 and 29 references (one appears in both sections).1
cation is proposed by Syntetos and Boylan [31] and Boylan et al.
We also analyze how these papers relate these aspects with stock
[32], through a two-dimensional matrix based on demand varia-
control (in Section 2.3). In the vast area of spare parts inventory
bility (cv2) and order frequency, while Williams [18] and Eaves
management modeling, instead, readers may resort to previous
and Kingsman [30] propose the partitioning of demand variance
literature reviews, such as [17]. Finally, seven papers discussed in
during the lead time. Yamashina [33] proposes the definition of
Section 2.4 deal with the gap between research and practice for
product-still-in-use quantity curves and service part demand
spare parts management.
curves as inputs for spare parts classification. Nagarur et al. [39]
and Porras and Dekker [40], instead, propose a hierarchical two- or
three-dimensional quali-quantitative classification. Ernst and
2.1. Spare parts classification
Cohen [28] apply the Operation Related Groups methodology,
based on a clustering technique, on 40 different variables, and
As pointed out by Boylan and Syntetos [10], spare parts for
finally Petrovic et al. [41] and Petrovic and Petrovic [42] propose
consumer products are highly varied, with different costs, service
an expert system combining the determination of failure rates and
requirements and demand patterns. A classification of spare parts,
fuzzy logic.
therefore, is helpful to determine service requirements for differ-
On the other hand, qualitative methods try to assess the
ent spare parts classes, and for forecasting and stock control
importance of keeping spare parts in stock based on information
decisions.
on the specific usages of spares and on factors influencing their
Table 1 summarizes the literature on classification methods
management (costs, downtime, storage considerations, etc.). The
for spare parts. In the review are also included works referring
Vital, Essential, Desirable (VED) is a qualitative method based on
more generally to low demand items, and some of the most-
consultation with experts [43]. Despite its apparent simplicity,
known general approaches to item classification. In total we
structuring a VED analysis might be a difficult task, as its accom-
analyzed 25 papers, 18 of which have proposed classifications
plishment may suffer from the subjective judgements of users [37].
developed for, or applied to, spare parts. The others, [18–24],
In order to limit this problem, VED may be combined with a
dealt instead (also) with slow moving demand items.
systematic procedure for classifying spare parts. For example,
Most of the analyzed papers propose multi-criteria classifica-
Gajpal et al. [26] propose a VED classification model based on the
tion methods, which take into account various factors. Only four
use of the Analytic Hierarchic Process (AHP) procedure [44,45],
papers [25,18,26,27] propose mono-criterion methods. It is also
defining three groups of spare parts (vital, essential and desirable).
important to notice that some papers propose the classification
A similar approach is proposed by Sharaf and Helmy [34],
criteria as their main contribution (e.g. [18,28,26]); for some
who consider five criteria and four alternative modes for each
others, instead, classification criteria are defined only within
criterion, determining four groups of spare parts: vital, essential,
examples of application of a proposed classification methodology
important, desirable. More complex is the approach adopted
(e.g. [21,23,22,24,29]).
by Braglia et al. [46], in which the decision tree about the multiple
Table 1 groups criteria into categories. The most popular
attributes analyzed (e.g. inventory constraints, lost production,
criteria, both appearing in 15 works, relate to part cost (unit or
logistic aspects) is integrated with a set of AHP models, in order
inventory cost), and part criticality. Other popular types of criteria
to solve the various multi-attribute decision sub-problems at the
are: demand volume or value (appearing in 13 contributions),
different nodes of the decision tree. Finally, Huiskonen [36] and
supply characteristics, such as replenishment lead time, supplier
Cavalieri et al. [37] do not propose a specific technique for
availability and risk of non-supply (indicated by 12 studies),
classifying spare parts according to the proposed criteria.
and demand variability, reported in 8 contributions—three of
Table 1 also reports about the empirical applications of the
which include the measure of lead time variability [30–32]. Other
proposed criteria and techniques. Only seven works [25,37,
criteria, proposed by fewer studies concern the part life cycle
30,32,40,35,27] report classification methods that have been actu-
phase [33–35], specificity [36,37], and reliability [33].
ally implemented by companies. Some other papers report a case
study as well, but just as a test of the proposed methodology
1
The main sources the of references reviewed in this section are the following [18,19,28,26,34,20,46,37,24,29], while some others [31,21,22,23,36]
journals: International Journal of Production Economics, Journal of the Operational present only a simulated quantitative or qualitative application on
Research Society, European Journal of Operational Research, International Journal real or hypothetical data. Four works [38,33,41,42], finally, do not
of Forecasting, International Journal of Operations and Production Management,
Management Science, Computers and Industrial Engineering, Computers & Opera-
report any application case. Eventually, aspects and obstacles related
tions Research, Production Planning and Control, Journal of Operations Manage- to the practical applicability of the classification methods, such as
ment and others. data availability, implementation algorithms, classification update,
 724

Table 1
Overview of the main spare parts classification contributes.

Reference Mono Multi- Classification criteria employed Classification technique: Application case
number criteria criteria
Quantitative Qualitative Case study with Case study Simulation Absent
implementation as a test
Part cost/ Part Supply characteristics/ Demand Demand Others ABC Other AHP Other
value criticality uncertainty volume/value variability

[25] X X X X
[18] X X X X
[38] X X X X X
[19] X X X X X
X X X Part reliability, Life X X
[33] cycle phase
[28] X X X X X X X
X X X X X Part weight, Repair X X
[42]
efficiency
[26] X X X X
[39] X X X X X
[36] X X X X X Part specificity X X
[34] X X X X X X Life cycle phase X X
[20] X X X X X X
[46] X X X X X X X
[30] X X X X X
[31] X X X X
[21] X X X X X X X
X X X X X X X
A. Bacchetti, N. Saccani / Omega 40 (2012) 722–737

[22]
[23] X X X X X X
Author's personal copy

[32] X X X X
[37] X X X X X X Part specificity X
[24] X X X X X X
[29] X X X X X
[40] X X X X X X
[27] X X X X
[35] X X X X X X Life cycle phase X X
 Author's personal copy

A. Bacchetti, N. Saccani / Omega 40 (2012) 722–737 725

solution sensitivity to thresholds or the role of judgment, are
discussed only by Huiskonen [36], Boylan et al. [32], and Syntetos
et al. [27].
In summary, following Syntetos et al. [27], we believe that the
issue of classification has not received as much academic attention
as its implications on spare parts management would require,
although it has increased in the last decade, when 16 of the 25
analyzed papers were published. In particular, there is a strong need
for case studies describing real implementation of classification
methods and focused on their practical applicability problems.
2.2. Spare parts demand forecasting
Fast moving parts may not require ad hoc forecasting methods,
but a large portion of spare parts are characterized by intermit-
tent or lumpy demand [10], requesting special attention. More-
over, demand for spares may be related to some explanatory
variables (e.g. failure occurrence, maintenance activities). Despite
the importance of demand forecasting for stock control, however,
spare parts research has mostly focused on inventory modeling
[47]. Only recently demand forecasting for spare parts (or more
generally for lumpy demand items) has been the object of
renewed research attention, and only one study to date carries
out a review of spare parts forecasting literature [4]. It provides a
useful framework to classify contributions according to the three
phases of forecasting: pre-processing (i.e. rules that designate a
spare part as slow or fast-moving, intermittent or lumpy);
processing (the application of the appropriate forecasting
method) and post-processing (adjustments made by the forecast
user). This section provides an extension of the review in [4],
focusing on the processing phase (most of the papers classified in
Section 2.1 belong, instead, to the pre-processing phase). Given
the relatively little specific attention devoted to spare parts, also
some relevant work on intermittent demand forecasting is con-
sidered. A summary of the reviewed works is reported in Table 2,
which groups literature according to the kind of method devel-
oped. Table 2 lists 28 contributions, 16 of which are from the last
decade. If we refer only to papers specifically addressing spare
parts or applying methods to spare parts demand forecasting, we
find out 13 works, 12 of which are from the years 2000s.
Time series demand forecasting methods have been applied to
spare parts. Traditional time-series methods, such as moving
average [48] or single exponential smoothing [49,50] are still
the most used in business practice. However, they have been
shown to overestimate the mean level of intermittent demand, if
applied immediately after a demand occurrence [4]. Johnston and
Boylan [51] proposed an adjusted exponentially weighted moving
average method (EWMA method) for forecasting intermittent
demand. They show that it performs better (in terms of mean
square of the forecast errors) than a traditional EMWA for an
inter-order interval higher than 1.25 forecasting periods.
Altay et al. [52] and Bermudez et al. [53], instead, developed
modifications to the Holt and Holt–Winters methods, respec-
tively, for intermittent demand in presence of trend or trend and
seasonality. In particular, Altay et al. [52] apply a modification of
Holt’s double exponential smoothing method developed by
Wright [54] to forecast intermittent demand with trend, both in
a simulated environment and with a real data set of aircraft parts
demand. Metrics on inventory levels and service implications
show that their method presents higher service levels but also
higher inventories than other specific forecasting methods (in
particular the one by Syntetos and Boylan [7], described later),
with no significant difference in the total cost.
A seminal work in the field of intermittent demand forecasting
is the one by Croston [6]: he suggested a method forecasting
separately, through single exponential smoothing, the interval
between demand arrivals and the demand size (see also [55,56]).
Syntetos and Boylan [57] showed that Croston’s estimator is
biased and proposed an adjusted method, showing its superior
performance [7]. Leven and Segerstedt [58] use the Croston
approach of only updating when there is a positive demand, but
update the forecast for the demand per period directly using the
ratio of demand size and interval. Their method, nonetheless, was
demonstrated being biased as well [59].
Bootstrapping was also applied to forecasting intermittent
demand [60,61]. Bootstrapping methods do not require to make
an assumption on the distribution of demand. The method by
Willemain et al. [61] modifies the traditional bootstrapping
approach [62] in order to take into account autocorrelation,
frequent repeated values and relatively short series, aspects often
occurring in spare parts demand. Using nine industrial datasets,

Table 2
Literature review of forecasting methods for lumpy demand.

Class Forecasting method Reference Classification Application to

number spare parts
Time Explanatory Hybrid Other forecasting
series

Traditional time series Traditional time series: moving average. exponential [49,50,48] X X
smoothing, EWMA
Modified traditional time Adjusted EWMA [51] X
series Adjusted Holt and Holt-Winters methods [53,52] X X [47]
Croston’s method and Croston [6,55,56] X
modifications Syntetos and Boylan approximation [57,7] X X
Croston modified [58] X
Bootstrapping Modified bootstrap method [60,61] X X
Demand aggregation/ Filtering/clustering methods X X
disaggregation [64,65]
Advance demand Order Overplanning [66,69] X
information Early sales [69,67,68] X
Analysis of reliability Failure rate analysis [33,76] X X
Operating condition analysis [75] X X
Regression Integrated Forecasting Method [74] X X
Neural networks Neural networks [71] X
Support vector machines [72,47] X X
Bayesian approach Beta-binomial model (allowing for different demand X
[77]
variance)
 Author's personal copy
726 A. Bacchetti, N. Saccani / Omega 40 (2012) 722–737
they compare their method with Croston’s and exponential
smoothing, achieving a higher forecasting accuracy. Although
their results received some criticism [63], bootstrapping repre-
sents an interesting alternative to Croston’s and its variants,
especially in cases in which short data history limit the reliability
of time series methods.
Other works focus on filtering demand data: Kalchschmidt
et al. [64,65] propose disaggregate time series, in order to
separate a stable demand series from an irregular (and lumpy)
one. To the different demand series thus generated, different
forecasting methods could be applied. In the spare parts case
discussed in [64] the authors apply a single exponential smooth-
ing to the stable demand series and a modified Croston’s proce-
dure to the lumpy one, obtaining reduced inventory levels, for
given service level objectives. A contribution of [64] stands in the
consideration of the supply chain structure and the customers’
characteristics as determinants of demand patterns and therefore
of the forecasting method to be adopted. The case study con-
siders, moreover, a quite common situation in durable consumer
goods, in which the product manufacturer provides spare parts to
a large network of small customers (e.g. repair shops) and a
smaller amount of large customers, such as regional warehouses,
subsidiaries and wholesalers.
Kalchschmidt et al. [64] also treat the value of advance
demand information from customers. This issue is modeled in
lumpy contexts by Verganti [66] through the order overplanning
model, which considers early information collected from custo-
mers as a driver for demand forecasts. Another method exploiting
early information is Early Sales [67,68]: in this case, orders from
customers with longer lead times are used to make a forecast of
the remaining demand. The application of this method to lumpy
demand is provided in [69]. These two methods, however, seems
ill-suited for forecasting spare parts requirements in durable
consumer goods industries. Order overplanning, in fact, requires
a knowledge and the collection of advanced information from a
large amount of customers in order to define the probability of
receiving orders. This seems unlikely in contexts with numerous
customers and items, and in which demand may be triggered by
unexpected events (product failures). On the other hand, the Early
Sales method requires a portion of customers to place orders in
advance, and those requests from early buyers to be correlated
with the other customers. These hypotheses may not be suited for
spare parts, where deliveries are generally requested urgently,
and the demand correlation among customers cannot be given for
granted, given the nature of events triggering demand. Never-
theless, improving forecasts and/or logistic performance through
buyer–supplier information exchange is a relevant issue also for
spare parts. [4] acknowledge a potential for savings related to
information sharing, but highlight the need for more research in
this area. According to our review, no study treated this issue for
spare parts demand, except for [64,70].
An alternative forecasting approach that has been applied to
spare parts is the one of neural networks. Gutierrez et al. [71]
apply a neural network model to 24 time series showing lumpy
demand patterns, and their method performed better than single
exponential smoothing, Croston’s method and the Syntetos–
Boylan approximation. Nonetheless, when a significant decrease
in the average of nonzero demand arises between the training
sample and the test sample, time-series methods tend to perform
better than neural networks. Hua and Zhang [47] propose an
application of the support vector machines (SVM) regression
method for forecasting spare parts demand (see also [72]). A
description of SVM mechanism is reported in [73]. [47] adopt a
hybrid logistic regression and the SVM approach aiming firstly to
forecast the occurrence of nonzero demands, and then to estimate
lead-time demand. Their test over a data set of spare parts
demand from a petrochemical company in China suggested this
method to perform better than Croston’s, bootstrapping and the
method by Hua et al. [74] (described later). A shortcoming of
neural network models, however, is the large amount of training
data required.
[74] propose a two-step integrated method for spare parts
forecasting in process industries. First, the occurrence of a
demand for spare parts is estimated as a function of some
explanatory variables (they consider plant and equipment over-
haul arrangements); the numerical value of demand, then, is
assessed through a bootstrapping method. They applied the
method to a data set of 40 spare parts from a petrochemical
enterprise in China, finding better performance compared to
exponential smoothing, Croston and a simple bootstrapping
method. The model developed by Ghodrati and Kumar [75],
instead, is purely explanatory. To a reliability analysis (to esti-
mate the failure rate of a component) they add a measure of the
impact of the working environment and the operators’ skills on
spare parts demand. Through a case study, it is shown that
including environmental factors in the calculation leads to more
accurate forecasts. Unfortunately, very specific information is
needed by explanatory methods, as shown by the applications
in [74,75]. This information is rarely available, except when the
supplier is in charge of maintenance activities at the customer’s,
or when the customer itself forecasts its spare part requirements.
Tibben-Lemke and Amato [76] proposed a model in which the
failure function of a part (normal wear) is coupled with a prediction
of future sales of the related end product in order to determine the
amount of replacement parts that will be demanded. The paper also
carries out a cost–benefit analysis of collecting information to
improve forecasts (in this case knowledge of product failure rates).
Yamashina [33], instead, based his model on the calculation of
product-still-in-use quantity curves and service part demand
curves. Finally, Dolgui and Pashkevich [77] developed a beta-
binomial model based on the Bayesian paradigm, which relies on
a beta distribution as a prior for the request probability of a group
of related stock-keeping units (SKUs) and allows for different
demand variance among the SKUs. The model is developed for
items with a short demand history.
Given the recent advances in spare parts demand forecasting
research, few established results emerge. In the time series
stream, most comparative studies supported the superiority of
Croston’s method and its variants over traditional time series
methods (see for instance, [78]). In particular, the Syntetos–
Boylan Approximation (SBA [57]) seems the best performing
one according to some studies [57,79], Teunter and Sani [80]
instead, suggest that another modification of Croston’s method by
Syntetos [81] overperforms the SBA.
Despite this evidence, there is still no conclusive and practi-
tioner-oriented indication on which is ‘‘the best’’ forecasting
method for spare parts. First of all, in fact, some studies support
the superiority of alternative methods to Croston’s or its variants
[61,74,71,52]: no comprehensive testing on a relevant amount of
data set was performed to give a conclusive answer on the issue.
Second, as [82] point out, results of comparative studies may be
biased by the usage of per-period forecast accuracy measures, not
suitable for series with many zero demand periods. Inventory and
service level measures are suggested as more suited. Nonetheless,
the adoption of different metrics in different studies may hamper
comparison. Third, most papers propose and evaluate techniques
given a generic context of lumpy demand, and only very few
propose criteria to differentiate the forecasting methods for
different items. Syntetos et al. [31] associate methods to items
according to a demand-based classification: the Croston forecast-
ing method (or its variants) should be applied to Intermittent,
Erratic and Lumpy items, the SES method to Smooth ones.
 Author's personal copy
A. Bacchetti, N. Saccani / Omega 40 (2012) 722–737
Bartezzaghi et al. [69] relate the method suitability to the
prevalent source of lumpiness. Kalchschmidt et al. [64] separate
stable and irregular demand series through customer disaggrega-
tion (filtering), allowing for differentiated techniques, Johnston
and Boylan [51] define a threshold of the order interarrival period
for which their method is superior to standard EWMA. According
to Syntetos and Boylan [10], finally, causal methods (relating
demand to other variables) can be used in the initial life cycle
phase of a part (since historical data are not available) or when
maintenance activities are under control of the vendor or a
business customer collecting historical data on some explanatory
variables. However, the role of spare parts classification criteria
(e.g. the ones discussed in Section 2.1), or of industry-related
factors on determining the most suited forecasting techniques
have not been thoroughly addressed by research.
Fourth, very few studies concerned the practical applicability
of methods to real cases for spare parts management. Require-
ments and obstacles to the implementation of forecasting meth-
ods in real cases, such as data availability, method complexity,
users skills, forecasting support systems and so forth have been
generally overlooked in academic research.
Eventually, authors tend to acknowledge that traditional single
exponential smoothing or moving average are still the most
commonly used methods by practitioners [10], and that specific
methods and techniques developed for spare parts are still
neglected in business practice. However, to our knowledge no
extensive empirical study about the diffusion of spare parts
forecasting practices in industrial contexts exist, while such
studies have been carried out for finished products sales fore-
casting (see [83] for a review). A partial exception is provided by
Cavalieri et al. [84]. In a preliminary study over a sample of 48
companies belonging to the automotive, household appliances
and consumer electronics industries, they found out that the
majority of companies did not forecast the demand for spare
parts, or relied only on judgemental methods. Only the auto-
motive sector had a majority of companies (58%) using a quanti-
tative forecasting method.
2.3. Integrating spare parts classification, demand forecasting with
stock control
Spare parts classification and demand forecasting can (and
should) be related to stock control policies [4]. Since the amount
of inventories due to slow-moving parts is generally important
[18,85,86]. even small improvements in the management of those
items may be translated into substantial cost savings [30].
We here consider the papers analyzed in Sections 2.1 and 2.2 that
also address inventory management issues. On the contrary, we
exclude from our analysis the quite vast amount of literature focusing
only on spare parts inventory management. About this topic, we
suggest to read seminal papers such as [87,88] or the literature
reviews by Guide and Srivastava [89] and Kennedy et al. [17].
Seven papers analyzed in Sections 2.1 and 2.2 [25,36,64,32,37,
40,27] develop a decision-making framework considering also
inventory management: only three of them [36,37,40], however,
do that at a general level. Gelders and Van Looy [25], Kalchschmidt
et al. [64], Boylan et al. [32] and Syntetos et al. [27] instead,
develop a framework for a specific case study analyzed, a general-
ization of which is not explicitly proposed. Moreover, some of
these works [36,33,44,90] limit their contributions to qualitative
guidelines for stock control. Finally, the methodologies by Duchessi
et al. [38], Petrovic and Petrovic [42] and Nagarur et al. [39] have
been implemented into computer programs supporting for spare
parts inventory control.
However, the papers quoted above suggest simple and well-
established inventory policies, mainly the continuous review
727
(Q, r), with fixed re-order point (r) and fixed order quantity
(Q) [25,32,40,27], and the continuous review (s, S), with fixed
re-order point (s) and order-up-to level (S) [38,39]. The latter is
considered the best-suited technique for low and intermittent
demand items by Sani and Kingsman [91]. The other analyzed
contributions give only qualitative guidelines on the inventory
policies to be adopted. As a matter of fact, as reported by Cohen
et al. [92] and Cavalieri et al. [37], few companies (in particular in
durable consumer goods industries) apply complex and specific
inventory models in practice, primarily due to the mathematical
complexity that characterizes their resolution.
This brief review, compared to the spare parts classification
and demand forecasting ones above, shows that in the scientific
literature an integrated perspective towards spare parts manage-
ment (classification, demand and inventory management alto-
gether) is often missing or limited. For instance, some of the cited
works remain at a too generic level with respect to practitioners’
requirements, while other are designed for a specific case study
and are based on a set of rules and/or algorithms tailored for that
specific situation, whose generalization to other settings could be
questionable.
2.4. The gap between theory and practice in spare parts
management
The existence of a gap between research and practice, in the field
of demand, inventory and production management has been sug-
gested by previous research [93,13,14,15]. Wagner [13], for instance,
observes that ‘‘despite half a century of impressive research on
inventory modeling y companies have poor customer service despite
excessive inventories’’ and that ‘‘incremental mathematical inventory
research is not likely to enhance practice’’.
Although no extensive studies have been carried out about spare
parts management practice, literature suggests that a distance exists
between research and practice in this field. Cohen and Agrawal [3],
for instance, state that: ‘‘Despite the aftermarket’s obvious charms,
however, most organizations squander its potential’’ (p. 130). Wagner
and Liedermann [16] studying seven machinery companies point
out: (i) a general lack of awareness and investment of time and
resources of top management in the spare part business, despite its
profit contribution, (ii) the little knowledge companies have of their
installed base; (iii) the little use of ad hoc inventory models for spare
parts, and (iv) even of the functionalities provided by ERP systems to
support forecasting and planning activities.
Syntetos et al. [27] observes that spare parts classification may
improve decision-making and constitutes a significant opportunity
for increasing spare parts availability and/or reducing inventory costs,
but despite its interest for practice, this issue has been overlooked in
the academic literature until recently. Boone et al. [11] identify
through a Delphi study the main challenges perceived by 18 senior
service part managers form different industries. The top three
challenges are: the lack of a system or holistic perspective, the
inaccuracy of service parts forecast, and the lack of system integration
among the supply chain parties. This highlights a strong need of
improving the integration of spare parts management processes both
at an internal and external level. This issue have been treated by
research about the design and management of supply chain and
networks (e.g. [94,95]), but little specific research on spare parts
supply chains exist. In addition, published case studies show that a
rigorous adoption of simple, non-optimized, classification and
inventory management methods is enough to provide great
benefits to organizations [27,39]. Finally, Zomerdijk and de
Vries [96] point out the importance of organizational context
for inventory management, criticizing the ‘‘one-sided’’ tradi-
tional operations research/operations management approach,
and develop an organizational perspective on inventory control
 Author's personal copy
728 A. Bacchetti, N. Saccani / Omega 40 (2012) 722–737
(considering also the task allocation, communication process,
decision making process and people behavior). Their frame-
work is applied to a spare parts management problem, under-
lining its potential benefits in this field.
The literature review in this section shows that theory about
spare parts classification, demand forecasting and integrated
spare parts management, despite its (especially recent) advance-
ments, presents some limitations, in particular in the light of its
practical applicability. The few empirical studies dealing with
spare parts suggest, in addition, the existence of a gap between
research and practice.
The rest of this paper is aimed at understanding how this gap
is expressed, and at identifying which directions for future
research may be drawn to help bridging this gap. This will be
done moving from an empirical perspective, as suggested by
Boone et al. [11], with specific reference to the durable goods
industries, and then by discussing the findings in the light of
previous research.
3. Research methodology
We carried out 10 case studies, with the aim of analyzing the
level of adoption of spare parts management practices. We chose
the cases on a conceptual ground, in order to have a representa-
tive sample. In fact, we analyze mass production manufacturing
companies providing durable goods addressed to private end
users or professional ones, rather than industrial users: the
sample companies belong to the automotive, digital printing,
heating & air conditioning, household and catering appliances
sectors. The case companies have an installed base ranging from a
few thousand to several million units, spread in different coun-
tries. In addition, these manufacturers manage after-sales net-
works mostly made by third parties providing technical
assistance and spare parts to final customers: the supply chain
structure is therefore multi-echelon, with a different number of
echelons co-existing in the same supply chain (as in the case
described by Kalchschmidt et al. [64]). The direct spare parts
customers are not the product end users, but the after-sales
network (dealers, repair shops, distributors, wholesalers, etc.).
The case companies, therefore, are characterized by a significant
supply chain complexity, a medium-to-high responsiveness
requirements in spare parts provision (not as high as in some
industrial contexts), and a rather wide range of stock keeping
units (SKUs) with differentiated demand patterns (slow and fast
moving, regular and lumpy). Finally, some of the considered
industries, such as the automotive and household appliance ones,
are renowned for their state-of-the-art operations management
practices (e.g. world class manufacturing, lean manufacturing): in
the sample we included some leading players in those industries,
that may be supposed to adopt ‘‘good’’ practices also concerning
spare parts management. The companies involved in the research
identified a key account for the study, in most cases the after-
sales logistics manager (at a corporate level, or of the Italian
subsidiary). The researchers held a preliminary interview with
this person, either by telephone or at the company premises,
lasting between 20 min and 1 h, to get an overview of the spare
parts range, the structure of the distribution network, and the
inventory and demand management practices. Then, a structured
questionnaire was sent to the company: the questionnaire and
the requested data were filled by the key account or other
company managers appointed by the key account. In general,
informants included the after-sales managing director and the
spare parts warehouse and material planning managers. The data
collection concerned: general information on the company, spare
parts description (number of codes, classification criteria,
volumes, etc.), distribution and customer networks, demand
forecasting and inventory management methods and processes,
supply chain information sharing and performance measurement
systems.
Once the questionnaire was filled, the researchers contacted
the companies for explanations and further details, also to
get a qualitative feeling on the studied subjects. E-mails, tele-
phone calls and/or face-to-face interviews with the personnel
involved allow to gather this supplementary information. More-
over, analysis of secondary sources (such as company documen-
tation and corporate websites) and direct observation (e.g.
warehouse tours, access to internal data when possible) were
used as supplementary sources of information, ensuring triangu-
lation of data.
The analysis consisted then of data reduction, data display
[97], and cross-case comparisons, to identify the main differences
and common behaviors among companies [98].
4. Empirical findings
4.1. Sample description
The sample companies belong to different industrial sectors:
automotive, household and catering appliances, digital printing,
heating and air conditioning. Moreover, they vary in size and
headquarter location, as shown in Table 3. Although a global
picture of the companies was obtained through the cases, most of
the collected evidence concern the Italian organization: six
companies are headquartered in Italy (A, B, C, D, E, L) and manage
subsidiaries abroad; in cases F, G, H, I, instead, we studied the
Italian subsidiary of a multinational company. In these cases,
however, Italy is an important market and the Italian after-sales
function is in charge of managing the third party after-sales
network, as well as spare parts inventories and distribution, at a
national level.
Table 4 reports the number of levels of the after-sales logistic
network directly owned or managed by the company, while
additional levels are generated by the customers’ warehouses.
Since most companies have mixed structures (e.g. some regions
are served directly by the central warehouse, some others
through a regional warehouse), Table 4 reports the highest
number of internal levels. Table 4 also reports the role of the
warehouse located in Italy and the number and types of custo-
mers it serves.
For all companies the majority of customers are repair centers
(on average 76% of the total). Repair centers tend to place small
orders very frequently and are in general served by express
deliveries. On the other hand, wholesalers, importers and sub-
sidiaries are larger customers, often ordering rather high amount
of parts with a lower frequency, sometimes creating sudden
demand peaks [64].
Table 5 describes the spare parts range managed by the case
companies. The average number of SKUs actually kept in stock is
about 17,000, with large differences within the sample. Companies
A, B, C, D, E, L with headquarter and central warehouse in Italy,
move on average 70% of their SKUs. Italian subsidiaries of foreign
firms (F, G, H, I) have, instead, a percentage of moved SKUs close to
100%, since the headquarters decided to allocate to the Italian
warehouse only ‘‘moving’’ SKUs. Finally, on average, 16% of SKUs
make 80% of ordered quantities, although with great differences
among companies. For companies D, E and G a very little portion
of SKUs (between 2 and 3%) accounts for 80% of volumes, while
on the contrary company A, dealing with professional users and
having a lower number of parts on stock, make the same volumes
with 46% of parts.
 Author's personal copy

A. Bacchetti, N. Saccani / Omega 40 (2012) 722–737 729

Table 3
An outline of the cases (company names are disguised for confidentiality reasons).

Company Headquarter Business Market and customers Size Turnover (Italy)

Italy Europe Extra White Heating & Automotive Printing Mainly Mainly Market Number of Company Percentage of
Europe goods Air systems professional private position employees turnover spare parts
conditioning end-users end- worldwide [Kh] turnover on
users [#] corporate
turnover

A X X X Niche player 243 22,600 8.8

B X X X Small player 135 21,600 13.9
C X X X Among the 250 45,500 1.4
main players
in the Italian
market
D X X X Relevant 7600 199,300 5.7
European
player
E X X X Among the 7200 192,000 7.9
main European
players
F X X X Premium 16,000 90,000 8.2
market leader
G X X X Small player 19,500 484,000 11.4
H X X X Premium 98,500 3,000,000 11.0
market Leader
I X X X Among the 170,000 155,000 8.1
main world
players
L X X X European 600 35,000 7.1
Leader in a
product
category

Table 4
Case companies logistics structure and customers.

Company Internal logistic Role of Italian Customers served by Italian warehouse(s)

network structure warehouse(s)
No. Repair centers (%) Wholesalers (%) Subsidiaries and
importers (%)

A 1 Level Central 530 81 1 18

B 2 Levels Central 1505 96 1 3
C 2 Levels Central 835 48 41 3
D 2 Levels Central 4446 83 16 1
E 2 Levels Central 3000 50 30 20
F 2 Levels Regional 287 100 0 0
G 3 Levels Regional þlocal 89 100 0 0
distribution centers
H 2 Levels Regional 220 100 0 0
I 2 Levelsa Regional 85 100 0 0
L 2 Levels Central 300 75 0 35

a
At the moment the case was carried out, the company was setting up a third level of Local Distribution Centers in some countries, among which Italy.

Table 5 4.2. Spare parts classification

Case companies spare parts range.
As shown in Table 6, six companies out of ten (A, B, D, E, F, I)
Company SKUs in stock Moved SKUs responsible
(previous year) of 80% of volumes (%)
adopt a simple mono-criterion classification method. Firm C,
instead, does not adopt any kind of spare parts categorization
A 1600 88% 46 for demand forecasting and stock control. It is noticeable that
B 2400 60% 20 G and H, companies belonging to the automotive industry, adopt
C 5300 68% 28
a two-criteria and a more structured classification approach than
D 37,700 55% 3
E 25,200 80% 2 the other sample firms. The most popular criterion is demand
F 7800 Almost 100% 20 volume (or value), adopted by eight firms, followed at distance by
G 19,300 Almost 100% 3 part cost (G, H) and part criticality (A, L). None of the other criteria
H 60,000 90% 15 pointed out in Section 2.1 is considered by the case companies.
I 3000 100% 22
L 10,000 76% 11
Among the companies that perform a classification of parts, A is the
only one that does not use the demand volumes or value, and relies
 Author's personal copy

730 A. Bacchetti, N. Saccani / Omega 40 (2012) 722–737

Table 6
Case companies spare parts classification.

Company Criteria Methodology No. of spare

parts classes
Demand Part Part No Quantitative (ABC) Quali-quantitative
volume/value cost/value criticality classification

A X X 3
B X X 3
C X 1
D X X 3
E X X 3
F X X 3
G X X X 9
H X X X 9
I X X 3
L X X X 6

Table 7
Importance of demand intermittence or lumpiness for the managed spare parts.

Consider demand intermittence or lumpiness phenomena: Comments

Very important (50% of SKUs or more)
Mildly relevant (less than 50% SKUs)
Not relevant (very low % of SKUs)
D, I, L The ‘‘gut feelings’’ of practitioner attribute to the majority of SKUs intermittent
or lumpy demand characteristics (more than 80% for company D)
B, C Company C was not able to estimate a % of codes characterized by lumpy or
intermittent demand
A, E, F, G, H F, G and H are subsidiaries of large multinationals, keeping on stock mainly
medium to fast moving SKUs
Company A manages a low number of SKUs, with no huge differences in
consumption (46% of SKUs make 80% of movements; see Table 5)
Company F shows an awareness problem

instead on a more qualitative definition of criticality classes, Table 8

although with little formalization of the classification process. This
also reflects the relative low number of SKUs and the relative even
volumes for the different SKUs, as commented earlier. The other
company not using a pure quantitative classification methodology is
L. Together with an ABC classification based on volumes, L adopts
the part criticality criterion (in a non-formalized way) to define
what class C codes to keep in stock, through qualitative judgement.
Case companies, in conclusion, prefer simple (quantitative)
classification methods like ABC, and consider no more than two
criteria: more articulated multi-criteria method described in
Section 2.1 are not adopted within the sample. More surprisingly,
firms seem not to carry out any specific evaluation of spare parts
demand characteristics at a classification level: elements such as
demand variation or order interarrival (see [31]) are not assessed.
As a consequence, the interviewed professionals answered to
questions about the degree of intermittence and lumpiness of
spare parts demand, reported in Table 7, through ‘‘gut feelings’’
rather than on the basis of quantitative analysis.
Partial exceptions are the Italian subsidiaries of multinational
companies, and in particular carmakers G and H, for which an
analysis of demand is carried out at the headquarters, and only
medium-to-fast moving codes are kept in stock at the Italian
warehouse. Slow-movers are instead delivered directly to custo-
mers from the central warehouse: in these cases a low perception
of the demand intermittence/lumpiness phenomena at the Italian
level seems more appropriate.
4.3. Spare parts demand forecasting
Table 8 shows that firm L does not carry out any kind of
forecasting: a pure reactive approach is adopted, with a periodic
review of stock levels. Company A, instead, generates forecasts on
a purely judgmental basis, and only for parts classified as critical.
Companies D and F adopt ‘‘black-box forecasting’’ approaches:
Case companies demand forecasting approaches.
Forecasting method Judgmental adjustments
Yes No
‘‘Black-box forecasting’’ D F
Traditional time series forecasting E, I, G, H B, C
Judgmental forecasting A
No forecasting L
forecasts are generated by an information system, but the specific
techniques are unknown to the users. Firm D users adjust the
forecasting output (the re-order proposals issued by the informa-
tion system in reason of demand forecasts) for ‘‘class A’’ SKUs.
Firm F (the Italian subsidiary of a foreign manufacturer), instead,
uses a forecasting, generated at the headquarters, without any
local judgmental adjustments.
The other case companies (B, C, E, G, H, I) carry out spare parts
demand forecasting with traditional time series methods (e.g.
Simple average, Moving average, Exponential smoothing). Company
E, in particular, generate forecasts only for ‘‘class A’’ SKUs (then
judgmentally adjusted). Firms B and C do not adjust the forecasts,
while G, H, I usually do it, even if in different ways. Carmakers G and
H receive the forecasts from a European headquarter that centralizes
the demand forecasting activity: Italian users may adjust them in
reason of their domain knowledge [99]. In the case of firm I, instead,
the Italian subsidiary directly generates the forecasts: adjustments
may be made on ‘‘class A’’ SKUs.
Although for five companies out of ten a significant amount of
spare parts is characterized by lumpy or intermittent demand
(see Table 7), no firm adopts techniques specific for intermittent
demand (e.g. Croston), because they prefer simpler methods, or
simply because they are not aware of the existence of specific
 Author's personal copy
A. Bacchetti, N. Saccani / Omega 40 (2012) 722–737
methods. Finally, we can point out that even in the nine companies
that carry out a spare parts categorization (see Table 6), there is
no differentiation in the forecasting techniques according to
the spare part class. Actually, the classification may determine only
if a forecasting activity has to be performed rather than how.
4.4. Spare parts stock control
Table 9 shows the re-order policies adopted by the sample
companies. A periodic review policy (T, R), is adopted by four
companies (A, F, D, L); firms B, G and H, instead, have a continuous
review with an order-up-to level (s, S), while a continuous review
with economic order quantity (Q, r) is adopted by C, E and I.
Seven companies out of ten keep safety stocks for spare parts:
for all SKUs (C, E, H, I), only for the most critical items (A) or for
slow moving items (F, L). Companies B, D and G instead do not
keep safety stocks.
The inventory management techniques used are rather simple
and well established, and not specifically developed for spare
parts. Moreover, as we previously noticed for demand forecasting,
inventory policies are rather undifferentiated among the different
spare part classes.
4.5. Downstream information sharing
In Table 10 we analyze the information sharing with down-
stream supply chain actors. We notice that almost half the sample
shows no information sharing on inventory levels or on demand
forecasts even with owned subsidiaries (in particular cases D and L).
Coordination with subsidiaries by companies B, E and I removes the
uncertainty about when subsidiaries will release orders, but not on
the ordered items and quantities. On the contrary, carmakers G and
H show a complete information integration between the head-
quarters and the regional warehouses and also with (at least a
significant share of) their authorized dealer network.
Moreover, dealers of companies G and H, through a web platform
can check item availability of SKUs unavailable at the manufacturer
from other dealers: this visibility encourages lateral transshipments
Table 9
Case companies stock control policies and safety stocks.
Inventory policy Safety stock
Yes No
Periodic review (T, R) A, F, L D carries out a categorization based on demand features. Also the
Continuous review (s, S) H B,G
Continuous review (Q, r) C, E, I
Table 10
Downstream Information sharing.
Information sharing Companies
No A,C, D,L
Yes, with company B,E, I: joint definition of re-order intervals
subsidiaries F: forecasts for the Italian market are carried out
at a central level
G, H: visibility of the stock at the regional (Italian)
level by the headquarters, centralized re-order
proposals
Yes, with the repair G, H: visibility of (part of) the dealers inventory
shops/dealer network level by the company
G, H: ‘‘lateral visibility’’: dealers can check the
availability of SKUs form other dealers (urgent
requirements and codes unavailable at the
company warehouse)
731
among dealers, improving both efficiency (dealers can reduce
inventories of these items) and effectiveness (improved customer
service).
5. Discussion
5.1. Summary of findings
Very few empirical studies compare spare parts management
practices of companies [16,11,92,84], and little research addressed
specifically the existence of a research–practice gap in spare parts
management. On the other hand, empirical works pointed out
that substantial benefits can be achieved through the adoption of
simple but formalized methods [27,39] or of an organizational
perspective towards spare parts inventory management [96].
The case studies show little (if any) adoption of ad-hoc
methods and techniques for spare parts management in the case
companies, the lack of integrated approaches (except for compa-
nies G and H), and a rather low level of awareness about how to
perform managerial improvements. Fig. 1 summarizes the case
findings, focusing on three of the main areas investigated: the
adoption of classification methods for spare parts, the adoption of
demand forecasting methods, and the information sharing within
the internal (between the parent company and subsidiaries/
warehouses) and external supply chain (with customers). Fig. 1
highlights the presence and specificity of the management tech-
niques used, as well as the degree of formalization (or specific
knowledge for spare parts management) emerging from the cases.
These findings support the idea of the existence of a significant
gap between research and practice for spare parts management.
Case companies, indeed, are rather representative of the fields of
durable goods addressed to private or professional users. In the
following, a detailed discussion of the case findings in the light of
extant theories is provided, leading to the identification of some
directions for future research to address this gap.
5.2. Analyzing the gap between research and practice
5.2.1. Obstacles to the adoption of advanced techniques and
methods
For spare parts classification, most of the case companies
resort to a traditional ABC method based only on demand volume
or value, and rarely consider a second criterion (criticality, cost),
in line with findings from Cohen et al. [92]. Notably, no company
actually implemented classifications reported in the literature
review in Section 2 are very simple, compared to some complex
methods developed by research. Similar evidence was collected
about inventory management and demand forecasting, for which
no company adopts techniques specifically developed for inter-
mittent demand. This is in line with previous empirical findings
[64,84,27]: practitioners, however, lament the inaccuracy of spare
Practice
adoption: Absent Traditional / low Advanced / specific
Area:
D F I L HG
Classification C
A B E
BC D F
Forecasting EG H I
L A
Supply chain C D
B E F I HG
integration L A
Fig. 1. Summary of findings from the case studies (the gray area represents cases
with little or no formalization, or no specific knowledge of the techniques used).
 Author's personal copy
732 A. Bacchetti, N. Saccani / Omega 40 (2012) 722–737
part demand forecasts [11]. Finally, the distribution of spare parts
in the case studies is based on multi-tier networks, in which at
least one intermediary exists between the manufacturer and the
final customers. The cases confirm findings from Boone et al. [11],
that point out the lack of a system perspective in the definition of
objectives, leading to sub-optimal decisions, and the low degree
of information sharing among suppliers, repairers, customers,
and service providers. Case companies lack an integrated view
of the supply chain, except for companies G and H, which set up
a vertical as well as horizontal visibility of inventories. This
integrated view is also scarce in the literature, as pointed out in
Section 2.3.
Some reasons can be found for the non-adoption of specific or
advanced models, and for the discrepancy between theory and
practice. First of all, in particular for demand forecasting, some
models are recent and not thoroughly tested, and unknown to
the interviewed professionals. More generally, however, models
proposed in literature are often perceived as too complex by
potential users [16]. Moreover, they may be considered too costly
to be implemented (investment in skills, human resources or
information support tools). In addition, decision-makers are keen
to base their decisions on experience and common sense [100].
Therefore managers prefer to rely only on their judgment or on
simple models (if they are needed for the large amount of SKUs to
be managed). Judgmental forecasting or judgmental adjustments
have been debated in research (see [99] for a literature review). It
is known that practitioners tend to use judgemental forecasting
as a correction factor of demand forecasts provided by statistical
methods, or as the sole forecast basis [101]. Goodwin [101],
resorting to previous studies, points out several reasons for that:
the perceived slowness of statistical extrapolative methods in
reacting to changes, the limited amount of available data, the
domain knowledge of human forecasters (e.g. knowledge of
events that are going to influence demand in the future), the lack
of skills, or the perceived loss of control and ownership when
forecasts are delegated to a statistical method. Finally, forecasts or
re-order proposals coming out from a method unknown or not
understood may be neglected by users [101,102]: this is the
‘‘black box’’ effect experienced by company D.
5.2.2. Analyzing the empirical findings from a contingency
perspective
Contingency theory provides a viewpoint for analyzing the gap
emerging from the case studies. The contingency perspective
[103,104] argues that differences in technological and environ-
mental dimensions result in differences in structure, strategies
and decision processes. Therefore, managerial effectiveness can
be achieved in more than one way (against a ‘‘one size fits all’’
view), but each way is not equally effective under all conditions
[105]: certain organizational actions are more appropriate than
others, depending on contextual factors [106]. The contingency
perspective has recently gained attention in the operations
management domain [107], as a natural evolution of the best
practice paradigm [108], and it looks at identifying under which
contextual conditions the adoption of operations management
practices is effective (see for instance [109]).
Through the contingency lens we analyze the level of adoption
of spare parts management practices by the case companies,
focusing on: scope of operations (domestic vs. international), firm
size (number of employees), strategic context (relevance of the
spare parts business), and industry. These contingency variables
allow to distinguish between contexts, and show variations in
the adoption of practices between the subsamples obtained.
Moreover, they are among the main categorical variables found
out in the literature review by Sousa and Voss [107]. Our
contingency analysis focus on the different degree of adoption
of spare parts management practices and tools, rather than on
their outcome (performances): thus, it can be classified as a
selection form of fit, according to Drazin and van de Ven [110].
Given the kind of research (qualitative, and on a low number of
cases compared to the one needed to gather statistical validity),
ours has to be considered an exploratory analysis. At an aggregate
level, however, it suggests significant differences in spare parts
management practices in different contexts. Below, the influence
of the different factors is discussed.
5.2.2.1. Scope of operations. Companies headquartered in Italy show
a lower degree of adoption of techniques and practices than
companies headquartered abroad. In particular, companies with
only domestic manufacturing activities (A, B, C, and L) are the ones
with the lowest level of practice adoption. Multinational companies
with international manufacturing operations, on the opposite, tend
to adopt more advanced management practices. This evidence may
be justified by the fact that companies with a broader scope of
operations have the cultural attitude, resources and tools that may
facilitate the process of adoption of spare parts management
practices, and it is supported by previous findings on operations
management practices (see [107]).
5.2.2.2. Size. All the companies with a lower adoption of practices
(A, B, C, L) have a small or medium size (from 135 to 600 employees),
while the rest of the sample has a much larger size (7000 or more
employees) and a greater adoption of practices. Size, therefore, seems
to affect the adoption of spare parts management practices, and this
can be explained by the different resource endowment of large versus
small firms, and is in line with findings from other contingency
research in operations management reviewed in [107].
5.2.2.3. Strategic relevance. Managerial literature stresses the role
of after-sales and spare parts business as a low-risk, long-term
revenue and profit source, a lever for differentiation, and a source of
knowledge on products and customers [3]. From a financial point of
view, all the companies declared spare parts margins significantly
higher than product margins. Spare parts make from 6% to 14% of
Italian revenues for all case companies, except C, which achieves
lower revenues (1.4%). Four companies (F, G, H, I) acknowledge a
strategic priority on after-sales activities: two of them (G and H) are
definitely adopting more advanced spare parts management
practices than the rest of the group, while the others (F and I)
stand in the middle. On the other hand, firms A, B and C seem to
hold onto the old vision of after-sales service as a ‘‘necessary evil’’
[111], and have a poor adoption of practices. More generally, we
may argue that all the case companies, to different degrees, are
living a transition phase in which higher importance and pressure is
put on spare parts management, an area rather overlooked in the
past, but their practices, skills and tools are still not adequate (with
the partial exception of G and H). This is in line with the findings
from Wagner and Liedermann [16].
5.2.2.4. Industry. Automotive companies (G and H) are a step ahead
the rest of the sample in managing the spare part business. These
companies show a coherent and integrated managerial approach,
although they do not resort to sophisticated techniques. This seems
consistent with the importance of after-sales to revenue, profit, and
competition in the automotive industry [112]. On the other hand,
the differences among the other firms seem not to be due to the
industry, but rather to the other factors described above. This
evidence is in line with a general higher degree of adoption of
operations management practices by automotive firms (e.g. lean
manufacturing) compared to other industries.
 Author's personal copy

A. Bacchetti, N. Saccani / Omega 40 (2012) 722–737 733

5.2.3. Analyzing the empirical findings from a knowledge maturity
perspective
A second approach to deepen the analysis of the empirical findings
is through the lens of a knowledge maturity model. This approach is
suggested by previous research on inventory management [15].
Actually, the results from the spare parts cases reflect a more general
problem of adoption of techniques and tools and of development of
organizational processes, that take place gradually with companies as
knowledge accumulates within the organization. In Fig. 2, the findings
from the cases are categorized into a knowledge maturity model,
adapting the maturity stages and management attributes proposed
by Niemi et al. [15], based on a previous work [113].
The framework in Fig. 2 identifies five maturity stages.
According to Niemi et al. [15], at the Initial stage the organization
possesses no specific knowledge on the subject (in our case spare
parts management); tacit knowledge and assumptions on basic
causalities are instead present at the Awareness stage. Knowledge
is made explicit at the Establishment stage, causalities are
expressed quantitatively at the Quantitative Management stage,
and they are finally thoroughly understood at the Optimization
stage. The three attributes represented in Fig. 2 are: the technical
tools (planning tools and operative data systems) used to carry
out the spare parts classification, demand and inventory manage-
ment tasks, the skills, roles and responsibilities of the operators
carrying out those tasks, and the performance measurement and
incentive systems related to spare parts management.
More than half the companies have only local planning tools
with no advanced functions (B, D, I) or tools are even absent
(A and L for forecasting, C for inventory). E and F use the company
ERP systems for basic planning functions, while G and H adopt ad-
hoc planning tools linked to the ERP system. The specific skills for
spare parts management appear to be deficient or basic in many
companies (i.e. some companies have little knowledge about the
lumpiness phenomenon and most do not know about the exis-
tence of specific forecasting methods for spare parts). In compa-
nies A, B, C, L the roles involved are those of operators (see [15]),
mainly dealing with day-by-day activities (control of deliveries
and ordering processes) with low emphasis on planning activities.
The operator task is supported by an expert in companies D, E, I
(at the Italian level), and F (at the European level). For companies
G and H, expert roles at the European level define the policies
followed by operators at the Italian level. Finally, the focus on
measurement systems is varied, as illustrated by Fig. 2, with some
kind of incentive systems in companies F, D, G, H and I.
From Fig. 2 it is possible to infer a low maturity for companies
A, B, C and L, corresponding to an ‘‘awareness’’ stage. Companies
D, E, F and I, although through different paths and with different
weaknesses, are moving into an establishment of their spare parts
management practice. Only companies G and H lean towards a
higher maturity of the spare part management process (quantita-
tive management), although ad hoc and optimization techniques
are not yet in place.
6. Bridging the gap: directions for future research
Building on the previous discussion, we propose four main
directions for future research, aimed to bridge the gap between
research and practice in the area of spare parts management.
(1) Developing an integrated approach to spare parts management:
Some limitations of the models proposed in literature have
been pointed out in Sections 2 and 5. However, a major
problem stands in the paucity of research with an integrated
perspective on spare parts management. Wagner and Lieder-
mann [16] found out only two papers on operations strategy
and design for spare parts management, contrasted to a
plethora of literature at the planning and operational levels.
Here we argue that an approach integrating spare parts
classification, demand management and forecasting, inven-
tory management models and performance measurement
is hardly present in literature (some partial examples are
reported in Section 2.3). We suggest that research develop
frameworks encompassing all these aspects. The adoption of
an integrated view, as depicted in Fig. 3, is one of the main
aspects affecting the overall effectiveness of spare parts
management in companies, leaving aside the complexity of
the specific techniques used; at the same time the absence of
an integrated view is one of the main gaps reported in the
Spare parts Demand Inventory Performance
classification forecasting management assessment
Fig. 3. An integrated approach to spare parts management.

Transaction Local tools ERP systems Advanced Advanced

systems local tools integrated tools

Technical tools A C L B D I E F HG

No skills, Basic skills, Controll skills, operator Advanced skills,

operator role operator role supported by expert manager role
Skills, roles and A B C L F I E D G H
responsibilities

No Aggregate, low- Focused measurement Diagnostic

measurement formalized measures and incentives measurement
Performance
measurement C A L B E F D I HG
and incentives

Maturity
stage Initial Awareness Establishment Quantitative Optimization
management

Fig. 2. Application of a knowledge maturity model to the case companies.

 Author's personal copy
734 A. Bacchetti, N. Saccani / Omega 40 (2012) 722–737
analyzed cases.The integrated perspective suggested in Fig. 3
stresses the relation between the steps of spare parts classi-
fication, demand forecasting and inventory management, and
the subsequent performance measurement. Decisions on
these aspects should be made with a systemic perspective,
and with a differentiated approach, where different kinds of
parts (according to the classification step) are treated with
different demand and inventory management techniques.
(2) Developing contingency-based managerial guidelines: To our
knowledge, no contingency research about spare parts man-
agement practices has been carried out. Extensive research
should be undertaken in this perspective. It may serve
different purposes, in line with the different approaches to
contingency research [107,110]: (a) to verify, at a single
practice level, the relation between contextual factors, such
as the ones analyzed in Section 5.2.2, and the adoption of
spare parts management practices (selection approach); (b) to
study how the interaction between contextual factors and
spare parts management practices affects performances, for
single factors (interaction approach) or with a holistic per-
spective (system approach); (c) to derive normative indica-
tions and guidelines for company managers, for instance
indicating the practices that can bring substantial benefits
under specific conditions as in [109].The analysis carried out
in this paper is not sufficient to give normative indications to
company managers. However, our research pointed out, for
instance, that firms from other industries mass-producing
goods may look at practices in the automotive sector as a
reference point. As well, confirming previous findings [3,16],
strategic commitment appears as a fundamental pre-condi-
tion to improve spare parts management. Finally, small firms,
not surprisingly, show a low adoption of spare parts manage-
ment practices. The definition of a development path for small
firms on how to gather the necessary skills (e.g. by co-
operating with non-competitive firms operating in the same
or similar industries) and to adopt the best suited practices
for their context should be the aim of future studies.
(3) Favouring the knowledge accumulation process: As pointed out
by Zomerdijk and de Vries [96], the organizational aspect is of
utmost importance for the success of the implementation of
management techniques in companies. Müller-Merbach [114]
reckons that scholars in operations research and management
science tend to neglect the implied process, i.e. the approach
to reality. Therefore, while adopting an integrated approach
and having identified which practices are more effective
under given conditions, researchers (and practitioners as
well) should ask themselves how this translates into deci-
sion-making processes, task allocation, communication pro-
cesses and how it is affected by personal behaviors of people
involved [96].Tools such as the knowledge maturity model
introduced in Section 5.2.3 can be an important contribution
of research to managerial practice. Efforts should be devoted
to further refining such knowledge maturity model for the
specificity of the spare parts field, in order to provide
companies with an instrument to understand their current
state of maturity in spare parts management. As well, general
transition paths for the different managerial attributes should
be developed. In addition, such kind of models may serve as
diagnostic tools to identify the weaknesses and prioritize
actions for the development of management attributes in real
companies (e.g. skills development, IT systems adoption,
etc.).Finally, such a research stream should investigate the
relations between different management attributes, drawing
inspiration from works in other operations management
fields and grounding concepts on in-depth case studies. For
instance, the relation between IT tools for spare parts
management and operators’ skills and responsibilities, as well
as the way in which performance and incentive systems may
facilitate the increase of knowledge into the organizations are
topics worth research interest.
(4) Supplementing models with practical relevance: Despite some
criticism emerging from this discussion, we believe that
research must continue to pursuit innovative, improved, and
sometimes complex quantitative models for spare parts man-
agement. Nonetheless, a great effort should be devoted to
evaluate how these models can be taken into practice. So,
researchers should focus on the assessment of the benefits
achievable by organizations applying such models [115]. As
well, the requirements and conditions for their practical applic-
ability, such as data availability, users’ skills, information sup-
port systems and so forth, have to be addressed. We encourage
the use of the case study method for these purposes, as well as
to facilitate the transferability of the developed solutions.
In addition, the impact of (upstream and downstream) informa-
tion sharing (e.g. about demand and inventories) and the organiza-
tional challenges for its implementation should be addressed, with
particular reference to the supply chain structures depicted by the
case studies in this paper: multi-echelon, with a large and differ-
entiated customer base. In such contexts, as for global operations,
the challenges of supply chain integration are great [116], but
worth the effort since they can improve supply chain performance
[95,117]. We already noted in the literature review that the way
information sharing affects demand forecasting accuracy is a topic
under-investigated by research. As pointed out by Martin et al. [70],
also the way local advance information (e.g. about product usage
conditions) can be exploited in order to improve planning decisions
should be addressed.
7. Conclusion
Despite the increasingly relevant role of spare parts management
in durable goods industries and its related challenges [3], the
research and business communities did not devote great attention
to the issue until recently. Very few studies, indeed, focus on
reviewing the state of the art of research advancements in this field
[17,89,4], or on analyzing managerial practices [92] and the dis-
crepancies with research [16]. The contribution of this paper moves
along these two directions, reviewing research on spare parts and
carrying out and discussing a multiple case study. As a third element
of contribution, the paper provides indications for future research.
First of all, a comprehensive literature review on research
about spare parts categorization, spare parts demand forecasting,
and their integration with inventory management approaches
was undertaken. The review in Section 2 adds to the existing
knowledge since it provides a systematic analysis of the research
contributions on these topics, identifying some research gaps,
such as the little attention about when the proposed methods are
superior to others and should be applied, and the lack of an
integrated perspective to the problem.
Secondly, we performed ten case studies in various sectors
(automotive, household appliances, printing systems, heating and
air conditioning). The empirical evidence supports the hypothesis
put forward by previous works [3,11,16] of a gap between the
models and methods developed by research and their actual
adoption in business practice. The analyses in Section 5 gave a
further contribution by exploring possible causes for this gap, at
three levels: the limitations of the proposed models in the
perspective of their practical adoption, the role of contingency
factors in explaining different levels of adoption of spare parts
 Author's personal copy
A. Bacchetti, N. Saccani / Omega 40 (2012) 722–737
management practices in the sample, and the level of maturity of
spare parts management in the studied organizations.
The third contribution of the paper stands in the identification
of four streams for future research in order to bridge this gap,
presented in Section 6. A major effort should go towards the
definition of research frameworks with an integrated perspective
on spare parts classification, demand forecasting, inventory man-
agement and performance measurement: such a perspective has
surprisingly been overlooked by both research [16] and practice
[11]. Such frameworks should allow to appreciate how manage-
ment practices or policies at one of these levels influence the
choices and their effects at other levels, and also to promote
the adoption of differentiated approaches (i.e. to different kind
of parts may correspond different management techniques).
A second indication for research concerns the application of the
contingency approach to spare parts management, in order to
appreciate the effectiveness of managerial practices in relation to
contextual factors, and to subsequently develop normative models.
Third, the refinement and adoption of knowledge maturity models
as diagnostic tools can indicate to companies their state of maturity
in spare parts management, and identify or prioritize development
actions. A fourth and final suggestion regards research on modeling
aspects (e.g. demand forecasting methods), which should increase
the emphasis on understanding the requirements for practical
applicability (e.g. data needs, required skills) and on assessing
the possible benefits of methods, resorting also to case study
applications.
Besides some merits, this research work presents important
limitations. Although it gave the opportunity to discuss all the
above issues, the set of case studied does not allow to generalize
our empirical findings: for that aim, a survey research is required.
Moreover, the sample cases present two other limitations: firstly,
the differences in size, industry and other firm characteristics,
although allowing to explore the effect of these aspects, may
hamper a direct comparison among the companies. Secondly and
more importantly, a true best practice case is missing. Although
this is itself a result of the empirical research (some of this
companies are leading in industries with quite advanced practices
in direct manufacturing and logistics), it also constitutes a serious
limitation of the paper, since we lack a reference point for
comparison (and for substantiating ‘‘advanced managerial prac-
tices’’). To overcome these limitations, more empirical research
should be carried out, as called by Syntetos et al. [27], both
through the survey and vertical case study methods, but this is
also part of the directions for future research presented in the
previous section of this paper.
Acknowledgements
The authors wish to thank the two anonymous referees for
their suggestions, leading to a considerable improvement of
the paper.
The study presented in this paper was carried out within the
ASAP Service Management Forum (www.asapsmf.org) a research
and dissemination initiative involving researchers from five
Italian universities, and several industrial and service companies.
References
[1] Gallagher T, Mitchke MD, Rogers MD. Profiting from spare parts.
The McKinsey Quarterly, 2005.
[2] Botter R, Fortuin L. Stocking strategy for service parts: a case study.
International Journal of Operation and Production Management 2000;20:
656–74.
[3] Cohen MA, Agrawal N. Winning in the aftermarket. Harvard Business
Review 2006;84:129–38.
735
[4] Boylan JE, Syntetos AA. Spare parts management: a review of forecasting
research and extensions. IMA Journal of Management Mathematics 2010;21:
227–37.
[5] Murthy DNP, Solem O, Roren T. Product warranty logistics: issues and
challenges. European Journal of Operations Research 2004;156:110–26.
[6] Croston. Forecasting and stock control for intermittent demands. Journal of
the Operational Research Society 1972;23:289–303.
[7] Syntetos AA, Boylan JE. The accuracy of intermittent demand estimates.
International Journal of Forecasting 2005;21:303–14.
[8] Aberdeen Group The service parts management solution selection report,
September 2005.
[9] Farris MT, Wittman CM, Hasty R. Aftermarket support and the supply chain.
International Journal of Physical Distribution and Logistics Management
2005;35:6–19.
[10] Boylan JE, Syntetos AA. Forecasting for inventory management of service
parts. Complex System Maintenance Handbook 2007;20.
[11] Boone CA, Craighead CW, Hanna JB. Critical challenges of inventory
management in service parts supply: a Delphi Study. Operation Manage-
ment Research 2008;1:31–9.
[12] Aberdeen Group Integrating spare parts planning with logistics, July 2008.
[13] Wagner HM. And then there were none. Operation Research 2002:50–217.
[14] Sanders NR, Manrodt KB. The efficacy of using judgmental versus quanti-
tative forecasting methods in practice. Omega 2003;31:511–22.
[15] Niemi P, Huiskonen J, Kerkkanen H. Understanding the knowledge accumulation
process—implications for the adoption of inventory management techniques.
International Journal of Production Economics 2009;118:160–7.
[16] Wagner SM, Liedermann E. A case study-based analysis of spare parts
management in the engineering industry. Production Planning & Control
2008;19:397–407.
[17] Kennedy WJ, Patterson JW, Fredendall LD. An overview of recent literature
on spare parts inventories. International Journal of Production Economics
2002;76:201–15.
[18] Williams TM. Stock control with sporadic and slow-moving demand.
Journal of the Operational Research Society 1984;35:939–48.
[19] Flores BE, Whybark DC. Implementing multicriteria ABC analysis. Engineer-
ing Costs and Production Economics 1988;15:191–5.
[20] Partovi FY, Anandarajan M. Classifying inventory using an artificial neural
network approach. Computers & Industrial Engineering 2002;41:389–404.
[21] Ramanathan R. ABC inventory classification with multiple-criteria using
weighted linear optimization. Computers & Operations Research 2006;33:
695–700.
[22] Zhou P, Fan L. A note on multi-criteria ABC inventory classification using
weighted linear optimization. European Journal of Operational Research
2006;182:1488–91.
[23] Ng WL. A simple classifier for multiple criteria ABC analysis. European
Journal of Operational Research 2007;177:344–53.
[24] Chen Y, Li KW, Kilgour DM, Hipel KW. A case-based distance model for
multiple criteria ABC analysis. Computers & Operations Research 2006;35:
776–96.
[25] Gelders L, Van Looy P. An inventory policy for slow and fast movers in a
petrochemical plant: a case study. Journal of the Operational Research
Society 1978;29:867–74.
[26] Gajpal PP, Ganesh LS, Rajendran C. Criticality analysis of spare parts using
the analytic hierarchy process. International Journal of Production Econom-
ics 1994;35:293–7.
[27] Syntetos AA, Keyes M, Babai MZ. Demand categorisation in a European
spare parts logistics network. International Journal of Operations & Produc-
tion Management 2009;29:292–316.
[28] Ernst R, Cohen MA. Operations related groups (ORGs): a clustering proce-
dure for production/inventory systems. Journal of Operations Management
1990;9:574–98.
[29] Chu C-W, Liang G-S, Liao C-T. Controlling inventory by combining ABC
analysis and fuzzy classifications. Computers & Industrial Engineering
2008;55:841–51.
[30] Eaves A, Kingsman B. Forecasting for the ordering and stock-holding of
spare parts. Journal of the Operational Research Society 2004;55:431–7.
[31] Syntetos AA, Boylan JE, Croston JD. On the categorization of demand
patterns. Journal of the Operational Research Society 2005;56:495–503.
[32] Boylan JE, Syntetos AA, Karakostas GC. Classification for forecasting and
stock control: a case study. Journal of the Operational Research Society
2008;59:473–81.
[33] Yamashina H. The service parts control problem. Engineering Costs and
Production Economics 1989;16:195–208.
[34] Sharaf MA, Helmy, HA. A classification model for inventory management of
spare parts. In: Proceedings of the 7th international conference on Produc-
tion, Industrial Engineering, Design and Control (PEDAC 2001), p. 375–82.
[35] Persson F, Saccani N. Managing the after sales logistic network—a simula-
tion study. Production Planning & Control 2009;20:125–34.
[36] Huiskonen J. Maintenance spare parts logistics: special characteristics and
strategic choices. International Journal of Production Economics 2001;71:
125–33.
[37] Cavalieri S, Garetti M, Macchi M, Pinto R. A decision-making framework
for managing maintenance spare parts. Production Planning & Control
2008;19:379–96.
[38] Duchessi P, Kumar Tayi G, Levy JB. A conceptual approach for managing
of spare parts. International Journal of Product Development 1988;18:8–15.
 Author's personal copy
736 A. Bacchetti, N. Saccani / Omega 40 (2012) 722–737
[39] Nagarur NN, Hu T, Baid K. A computer based inventory management system
for spare parts. Industrial Management & Data Systems 1994;94:22–8.
[40] Porras E, Dekker R. An inventory control system for spare parts at a refinery:
an empirical comparison of different re-order point methods. European
Journal of Operational Research 2008;184:101–32.
[41] Petrovic D, Petrovic R, Senborn A, Vujosevic A. A microcomputer expert
system for advising on stocks in spare parts inventory systems. Engineering
Costs and Production Economics 1990;19:365–70.
[42] Petrovic D, Petrovic R, SPARTA II. Further development in an expert system
for advising on stocks of spare parts. International Journal of Production
Economics 1992;24:291–300.
[43] Mukhopadhyay SK, Pathak K, Guddu K. Development of decision support
system for stock control at area level in mines, IE(I) Journal-MN, 2003.
[44] Saaty, TL. Multicriteria decision making: the analytic hierarchy process,
Revised and published by the author, Original version published by
McGraw-Hill, New York, 1980.
[45] Saaty TL. How to make a decision: the analytic hierarchy process. European
Journal of Operational Research 1990;48:9–26.
[46] Braglia M, Grassi A, Montanari R. Multi-attribute classification method for
spare parts inventory management. Journal of Quality in Maintenance
Engineering 2004;10:55–65.
[47] Hua H, Zhang B. A hybrid support vector machines and logistic regression
approach for forecasting intermittent demand of spare parts. Applied
Mathematics and Computation 2006;181:1035–48.
[48] Makridakis S, Wheelwright SC, Hyndman RJ. Forecasting: methods and
applications. 3rd ed New York: John Wiley & Sons, Inc.; 1998.
[49] Brown RG. Statistical forecasting for inventory control. New York: McGraw-
Hill; 1959.
[50] Brown RG. Smoothing forecasting and prediction of discrete time series.
New Jersey: Prentice-Hall; 1962.
[51] Johnston FR, Boylan JE. Forecasting for items with intermittent demand.
Journal of the Operational Research Society 1996;47:113–21.
[52] Altay N, Rudisill F, Litteral L. Adapting Wright’s modification of Holt’s
method to forecasting intermittent demand. International Journal of Pro-
duction Economic 2008;111:389–408.
[53] Bermudez JD, Segura JV, Vercher E. Improving demand forecasting accuracy
using nonlinear programming software. Journal of the Operational Research
Society 2006;57:94–100.
[54] Wright DJ. Forecasting data published at irregular time intervals using an
extension of Holt’s method. Management Science 1986;32:499–510.
[55] Rao AV. A comment on : forecasting and stock control for intermittent
demands. Operational Research Society 1973;24:639–40.
[56] Schultz CR. Forecasting and inventory control for sporadic demand under
periodic review. Journal of the Operational Research Society 1987;38:
453–8.
[57] Syntetos AA, Boylan JE. On the bias of intermittent demand estimates.
International Journal of Production Economics 2001;71:457–66.
[58] Leven E, Segerstedt A. Inventory control with a modified Croston procedure
and Erlang distribution. International Journal of Production Economics
2004;90:361–7.
[59] Boylan JE, Syntetos AA. The accuracy of a modified Croston procedure.
International Journal of Production Economics 2007;107:511–7.
[60] Snyder R. Forecasting sales of slow and fast moving inventories. European
Journal of Operational Research 2002;140:684–99.
[61] Willemain TR, Smart CN, Schwarz HF. A new approach to forecasting
intermittent demand for service parts inventories. International Journal of
Forecasting 2004;20:375–87.
[62] Efron B. Bootstrap methods: another look at the Jackknife. Annals of
Statistics 1979;7:1–26.
[63] Gardner ES, Koehler AB. Correspondence: comments on a patented boot-
strapping method for forecasting intermittent demand. International jour-
nal of forecasting 2005;21:617–8.
[64] Kalchschmidt M, Zotteri G, Verganti R. Inventory management in a multi-
echelon spare parts supply chain. International Journal of Production
Economics 2003;81–82:397–413.
[65] Kalchschmidt M, Verganti R, Zotteri G. Forecasting demand from hetero-
geneous customers. International Journal of Operations & Production
Management 2006;26:619–38.
[66] Verganti R. Order overplanning with uncertain lumpy demand: a simplified
theory. International Journal of Production Research 1997;35:3229–48.
[67] Fisher ML, Raman A. Reducing the cost of demand uncertainty through
accurate response to early sales. Report no. 92-10-03, Harvard University,
USA, 1992.
[68] Fisher ML, Hammond JH, Obermeyer WR, Raman A. Making supply meet
demand in an uncertain world. Harvard Business Review, 1994;May/
June:83–93.
[69] Bartezzaghi E, Verganti R, Zotteri G. A simulation framework for forecasting
uncertain lumpy demand. International Journal of Production Economics
1999;59:499–510.
[70] Martin H, Syntetos AA, Parodi A, Polychronakis YE, Pintelon L. Integrating
the spare parts supply chain: an inter-disciplinary account. Journal of
Manufacturing Technology Management 2010;21:226–45.
[71] Gutierrez RS, Solis AO, Mukhopadhyay S. Lumpy demand forecasting using
neural networks. International Journal of Production Economics 2008;111:
409–20.
[72] Bao Y, Wang W, Zhang J. Forecasting intermittent demand by SVMs
regression. In: IEE international conference on systems, man and cyber-
netics, p. 461–6.
[73] Vapnik VN. The nature of statistical learning theory. New York: Springer;
1995.
[74] Hua ZS, Zhang B, Yang J, Tan DS. A new approach of forecasting intermittent
demand for spare parts inventories in the process industries. Journal of the
Operational Research Society 2007;58:52–61.
[75] Ghodrati B, Kumar U. Operating environment-based spare parts forecasting
and logistics: a case study. International Journal of Logistics: Research and
Applications 2005;8:95–105.
[76] Tibben-Lemke RS, Amato HN. Replacement parts management: the value of
information. Journal of Business Logistics 2001;22:149–64.
[77] Dolgui A, Pashkevich M. Demand forecasting for multiple slow-moving
items with short requests history and unequal demand variance. Interna-
tional Journal of Production Economics 2008;112:885–94.
[78] Willemain TR, Smart CN, Shockor JH, DeSautels PH. Forecasting intermittent
demand in manufacturing: a comparative evaluation of Croston’s method.
International Journal of Forecasting 1994;10:529–38.
[79] Syntetos AA, Boylan JE. On the stock control performance of intermittent
demand estimators. International Journal of Production Economics 2006;103:
36–47.
[80] Teunter RH, Sani B. On the bias of Croston’s forecasting method. European
Journal of Operational Research 2009;194:177–83.
[81] Syntetos AA. Forecasting for intermittent demand. Unpublished PhD
thesis. Buckinghamshire Chilterns University College, Brunel University,
2001.
[82] Teunter RH, Duncan L. Forecasting intermittent demand: a comparative
study. Journal of the Operational Research Society 2009;60:321–9.
[83] Zotteri G, Kalchschmidt M. Forecasting practices: Empirical evidence and a
framework for research. International Journal of Production Economics
2007;108:84–99.
[84] Cavalieri S, Perona M, Pinto R, Saccani N. After Sales service in durable
consumer goods: the case of Italian industry. In: Proceedings of the 13th
annual international EurOMA conference 2006, p. 779–88.
[85] Dunsmuir WT, Snyder R. Control of inventories with intermittent demand.
European Journal of Operational Research 1989;40:16–21.
[86] Vereecke A, Verstraeten P. An inventory management model for an inventory
consisting of lumpy items, slow movers and fast movers. International Journal
of Production Economics, Elsevier 1994;35:379–89.
[87] Silver EA. Some ideas related to the inventory control of items having
erratic demand patterns, 1970, CORS, 8, 87100.
[88] Ward JB. Determining reorder points when demand is lumpy. Management
Science 1978;24:623–32.
[89] Guide DR, Srivastava R. Repairable inventory theory: models and applica-
tions. European Journal of Operational Research 1997;102:1–20.
[90] Dekker R, Bayindir ZP. Spare parts inventory control: an overview of issues
for a large industrial complex. IMS International Forum-2004 Global
Challenges in Manufacturing, p. 180–7.
[91] Sani B, Kingsman BG. Selecting the best periodic inventory control and
demand forecasting methods for low demand items. Journal of the Opera-
tional Research Society 1997;48:700–13.
[92] Cohen MA, Zheng Y, Agrawal V. Service parts logistics: a benchmark
analysis. IIE Transactions 1997;29:627–39.
[93] Van Donk DP. Make to stock or make to order: the decoupling point in the
food processing industries. International Journal of Production Economics
2001;69:297–306.
[94] Georgiadis MC, Tsiakis P, Longinidis P, Sofioglou MK. Optimal design of
supply chain networks under uncertain transient demand variations.
Omega 2011;39:254–72.
[95] Kim KK, Ryoo SY, Jung MD. Inter-organizational information systems
visibility in buyer–supplier relationships: the case of telecommunica-
tion equipment component manufacturing industry. Omega 2011;39:
667–76.
[96] Zomerdijk L, de Vries J. An organizational perspective on inventory control:
theory and a case study. International Journal of Production Economics
2003;81–82:173–83.
[97] Miles MB, Huberman AM. Qualitative data analysis: an expanded source-
book. 2nd ed London & Thousand Oaks, CA: Sage; 1994.
[98] Yin RK. Case study research—design and methods. 2nd ed. Thousand Oaks,
CA: Sage; 1994.
[99] Lawrence M, Goodwin P, O’Connor M, Onkal D. Judgmental forecasting:
a review of progress over the last 25 years. International Journal of
Forecasting 2006;22:493–518.
[100] Zsambok CE. Naturalistic decision-making: where are we now? In: Zsambok CE,
Klein G, editors. Naturalistic decision-making. Lawrence Erlbaum Associates;
1997. p. 3–16.
[101] Goodwin P. Integrating management judgment with statistical methods to
improve short-term forecasts. Omega 2002;30:127–35.
[102] Fransoo JC, Wiers VCS. An empirical investigation of the neglect of MRP
information by production planners. Production Planning & Control
2008;19:781–7.
[103] Woodward, J. Management and technology. London: H.M.S.O.; 1958.
[104] Thompson J. Organizations in action. New York: McGraw-Hill; 1967.
 Author's personal copy
A. Bacchetti, N. Saccani / Omega 40 (2012) 722–737
[105] Zeithaml VA, Varadarajan PR, Zeithaml CP. The contingency approach: its
foundations and relevance to theory building and research in marketing.
European Journal of Marketing 1988;22:37–64.
[106] Galbraith JR. Organisational design. Reading, MA: Addison-Wesley; 1977.
[107] Sousa R, Voss CA. Contingency research in operations management prac-
tices. Journal of Operations Management 2008;26:697–713.
[108] Voss CA. Alternative paradigms for manufacturing strategy. International
Journal of Operations & Production Management 1995;15:5–16.
[109] Olhager J, Prajogo DI. The impact of manufacturing and supply chain
improvement initiatives: a survey comparing make-to-order and make-to-
stock firms. Omega 2012;40(2):159–65.
[110] Drazin R, van de Ven A. Alternative forms of fit in contingency theory.
Administrative Science Quarterly 1985;30:514–39.
[111] Lele MM. After-sales service—necessary evil or strategic opportunity?
Managing Service Quality 1997;7:141–5.
737
[112] Cohen MA, Cull C, Lee HL, Willen D. Saturn’s supply-chain innovation: high
value in after-sales service. Sloan Management Review 2000:93–101.
[113] Paulzen O, Perc P. A maturity model for quality improvement in knowledge
management. In: Wenn A, McGrath M, Burstein F, editors. Enabling organiza-
tions and society through information systems, Proceedings of the 13th
Australasian Conference on Information Systems (ACIS2002), p. 243–53.
[114] Müller-Merbach H. The OR/MS process: an orphan? The five sections
analysis Omega 2010;38:115–7.
[115] Thonemann U, Brown A, Hausman W. Easy quantification of improved spare
parts inventory policies. Management Science 2002;48:1213–25.
[116] Gunasekaran A, Cheng TCE. Special issue on logistics: new perspectives and
challenges. Omega 2008;36:505–8.
[117] Trapero JR, Kourentzes N, Fildes R. Impact of information exchange on
supplier forecasting performance. Omega 2012;40(6):738–47.