ARTICLE IN PRESS

Int. J. Production Economics 118 (2009) 501507

Contents lists available at ScienceDirect

Int. J. Production Economics

journal homepage: www.elsevier.com/locate/ijpe

Trading off between heijunka and just-in-sequence

Andreas Huttmeir a, Suzanne de Treville a,, Ann van Ackere a, Leonard Monnier a,
Johann Prenninger b
a
b

Faculty of Business and Economics, University of Lausanne, 1015 Lausanne-Dorigny, Switzerland

BMW AG, Munich, Germany

a r t i c l e in fo

abstract

Article history:
Received 25 October 2006
Accepted 11 December 2008
On-line el 19 January 2009

The concept of heijunkacontrolling the variability of the job arrival sequence to permit
higher capacity utilizationplays an integral role in lean production theory. In
situations where the customer denes the delivery sequence, however, scheduling
production to maximize utilization becomes more challenging and requires a
subsequent reordering. The cost of the extra work and space required by this reordering
needs to be traded off against the value of the higher utilization. We present the results
of a stylized simulation-based model of the two approaches inspired by a BMW engine
plant.
& 2009 Elsevier B.V. All rights reserved.

Keywords:
Just in sequence
Heijunka
Lean production
Simulation
Scheduling

0. Introduction
Production systems must always balance inventory,
capacity utilization, and system variability according to
the laws of Factory Physics (Hopp and Spearman, 1996).
de Treville and Antonakis (2006) took the concept a step
further, proposing to dene lean production according to
its Factory Physics. Lean production thus dened calls for
high capacity utilization combined with relatively low
inventories, requiring that system variability (concerning
both arrival and service rates) be minimized. Many lean
production practices serve to either control inventory
buildup (kanban) or to reduce system variability (through
defects, unplanned downtime, exposure to worker absenteeism, etc.). The lean production practice that protects
the producer from variability in the sequence of jobs to be
processed is heijunka, in which production is scheduled
such that the production line produces the same sequence
of products throughout a given time period, with that
sequence alternating between demanding and less demanding products. The assumption underlying heijunka is

 Corresponding author. Tel./fax: +41 21 692 3341.

E-mail address: suzanne.detreville@unil.ch (S. de Treville).

0925-5273/$ - see front matter & 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.ijpe.2008.12.014

that the producer has a choice concerning the amount of

variability in the job arrival sequence to accept. Rather
than naively accepting all jobs that arrive in the order
received, the producer can choose to sort orders so that
the arrival sequence seen by production is relatively
smooth.
As product variety increases, however, the practice of
heijunka becomes more challenging. Also, eliminating
variability is not always strategically desirable: A supplying company can increase competitiveness through
being able to respond to variations in customer needs
(variability that can be exploited to gain a competitive
edge is referred to as strategic variability, Suri, 2003). An
alternative approach, referred to as just-in-sequence (JIS)
scheduling, has begun to play an increasing role in lean
production, even though production lines operating JIS
will clearly be less lean than those operating according to
heijunka. To what extent should a company be willing to
compromise its leanness to take advantage of such
strategic variability?
The choice between heijunka and JIS scheduling
illustrates the tension between leanness and agility that
increasingly must be sorted out if manufacturing is to be
competitive (Prince and Kay, 2003; Yao and Carlson, 2003;
Narasimhan et al., 2004; Browning and Heath, 2008). In

ARTICLE IN PRESS
502

ttmeir et al. / Int. J. Production Economics 118 (2009) 501507

A. Hu

this paper, we combine a stylized simulation model with a

case study of a BMW engine plant to gain insights into the
nature of the trade-off and explore ways to combine
limited heijunka with partial JIS production.
The paper is structured as follows: Sections 1 and 2
describe heijunka and JIS and provide a comprehensive
literature review. Section 3 describes the issues the BMW
engine plant was facing in 2003 as it worked through the
question of how to schedule production. The model and
its results are presented in Section 4. Finally, the main
ndings and implications are briey summarized in
Section 5.

1. Heijunka
The objective of heijunka is to avoid peaks and valleys
in the production schedule. Consider, for example, a
workstation that produces two products, A and B, with A
requiring 1.5 min, and B requiring 1 min of processing,
respectively. Suppose that the company receives an order
for 100 units of both A and B. A nave schedule would
be to produce 100 units of one product and then 100
units of the other, resulting in a situation in which the
demand faced by the workstation would vary considerably. Transport this workstation to a production line with
a cycle time of 1.4 min, and the workstation is overloaded
(and a bottleneck for the entire line) for 100 cycles and
underloaded for another 100 cycles. Accommodating this
schedule requires increasing the cycle time for the entire
line, at least during the period when A is being produced.
On many production lines it is not realistic to change the
cycle time to accommodate such workload uctuations
(this would not be possible at the BMW engine plant
described in this paper, for example), hence this workstation might well be obliged to operate at relatively low
average capacity utilization.
Heijunka calls for distributing the jobs requiring more
labor input throughout the production schedule to permit
higher average utilization assuming that the cycle time is
held constant over time. In our simple example, products
A and B would be alternated, so that the workstation
could either work in lots of one unit of A and one unit
of B, with cycle time determined based on the combined
work content of 1.5+1.0 2.5 min, or allowing the workstation to get a bit behind during the cycle when A is
produced, catching up during the cycle when B is
produced. Monden (1983) suggested a simple algorithm
for heijunka scheduling that has been used in practice.
Heijunka (also referred to as production smoothing or
leveling the production schedule) has played an integral
role in just-in-time and lean production since its inception
(Schonberger, 1982; Hall, 1983; Monden, 1983; Womack et
al., 1990; Chase, 1993; Hopp and Spearman, 1996;
Fujimoto, 1999). Abdulmalek and Rajgopal (2007), for
example, described use of heijunka (implemented using
Mondens algorithm) as one of several lean production
practices implemented on the cold end of an integrated
steel mill.
Teece et al. (1997) referred to heijunka as an example
of a dynamic manufacturing capability or routine that

might increase rm competitiveness when operating in

conguration with other routines or capabilities.
McLachlin (1997, p. 287) listed uniform plant loading
as a JIT ow element, operationalized as (a) there is a
xed and level schedule; (b) they produced the same mix
of end items or families each day, and possibly each hour
(to match daily demand rates); (c) there is a reduction in
upstream inventory swings and panic reactions to
(changes in) schedule(d) demand; and (d) there is little
or no expediting. In six JIT implementations evaluated
using a Likert scale ranging from 1 to 6, two rated their
level of uniform plant loading at 4, two at 3, one at 2, and
one at 1, indicating that for these plants leveling the
production schedule was not a top priority.
Shah and Ward (2003) did not include heijunka or
leveling the production schedule in their extensive list of
lean manufacturing practices. Production smoothing was
mentioned in the context of bottleneck removal rather
than schedule leveling, and there was a general reference
to planning and scheduling strategies, which could refer
to either heijunka or JIS scheduling.
Rinehart (1997) however, described difculties faced
by one lean auto assembly plant in leveling the production
schedule. Although heijunka was a stated company policy,
variation in demand meant that it was constantly violated.
Unfortunately, production was scheduled and capacity
allocated under the assumption that heijunka was
functioning correctly, so surges of overburdened jobs
resulted in extreme worker stress, repetitive strain
injuries, and potential quality problems. In labor disputes
that eventually emerged, overburdened jobs resulting
from a failure to maintain heijunka were a major point
of conict.
De Smet and Gelders (1998) noted that implementation of heijunka was only possible in situations where
there were few schedule disturbances, that is, demand
was relatively stable and predictable. Cusumano (1994) in
his article summarizing the limits to lean production
observed that production smoothing through techniques
like heijunka reduces manufacturing exibility (see also
Hines et al., 2004). As mentioned in the introduction,
other authors have called for an addition of agility to
leanness, indicating a need to explore the trade-off
between leanness and responsiveness (Prince and Kay,
2003; Yao and Carlson, 2003; Narasimhan et al., 2004;
Browning and Heath, 2008). Such a trade-off has implications for the implementation of heijunka. The purpose of
heijunka is to protect the production line from demand
volatility, but when that demand volatility is strategic,
there may be a need to relax or even eliminate heijunka. In
cases where manufacturing exibility must be added to
leanness, heijunka may be replaced by a JIS production
planning and control approach, which we describe in the
following section.

2. JIS production scheduling

JIS occurs according to a sequence dened by the
customer throughout the supply chain. It can be implemented in delivery, production, and distribution. Delivering

ARTICLE IN PRESS
ttmeir et al. / Int. J. Production Economics 118 (2009) 501507
A. Hu

JIS requires either an inventory of parts that can be picked

in the right sequence, or producing and/or assembling parts
in the order needed (Sayer, 1986; Kempfer, 2005). JIS
delivery is viewed as playing an increasing role in effective
manufacturing (Liker and Wu, 2000), and is now required
by an increasing number of companies in addition to BMW
such as DaimlerChrysler, Saturn, and Dell (Prenninger,
2001; Feare, 2003; Trebilcock, 2006).
One of the earliest success stories of JIS production and
delivery is told in the Toyota Motor Manufacturing case
(Mishina and Takeda, 1994), describing how the seat
supplier developed the capability to produce and deliver
seats in the order required. Toyota had originally suggested that the seat supplier produce to stock and deliver
JIS (as was done for all other Toyota plants). The seat
supplier argued, however, that the production and
transportation lead times for the seat were short enough
to permit seat production to begin after the car left
painting (the point at which the nal sequence was
established due to variability in painting operations) and
be delivered to the factory oor in time to be installed.
Also, the number of seat variations was sufciently
high to require a quite large warehouse, and the number
of variations was increasing over time. The seat supplier
was located 30 min away from the Toyota plant, making
JIS production easier. The case also describes, however,
some of the problems arising with JIS when quality
problems occurred, as JIS leaves much less latitude for
rework. JIS also forced the seat supplier and Toyota
to work together to x the quality problem much
more aggressively than would have occurred otherwise,
so that the disruptions led to organizational learning.
This is consistent with Narasimhan et al.s (2006) results
that showed that lean producers outperformed agile
producers in terms of cost, but not in terms of conformance quality. In other words, just as removal of
inventory has been demonstrated to lead to learning
under the right conditions (Suri and de Treville, 1986), so
JIS production can also create powerful motivation for
improvement.
Some authors have described use of a postponement
strategy to permit JIS delivery and assembly from a
decoupling inventory, allowing an emphasis on high
utilization upstream, and responsiveness downstream
(Feitzinger and Lee, 1997; Lambert et al., 1998; van Hoek,
2001; Abdulmalek and Rajgopal, 2007).

3. BMW
Along the lines discussed in the previous two sections,
BMW faced a choice between heijunka and JIS production
in one of their European engine factories in 2003. The
engine plant had historically been organized around the
concept of heijunka, permitting extremely stable production and a high utilization (ranging between 95% and
99%). The smoothing of the production schedule was
carried out by the advanced planner and optimizer (APO)
module of the SAP/R3 system. Engines were delivered to
the customer (the assembly plants) approximately in the
order in which they were produced.

503

During the late 1990s, BMW management moved

responsiveness to a top priority, implementing a company-wide initiative entitled customer-oriented sales
processing to achieve delivery of a car within 10 working
days of a customer placing an order (Selto et al., 1995;
Prenninger, 2001; Pietsch, 2002). At the same time,
strategic variability (variability which an organization
uses to maintain its competitive edge) was increasing,
such that today there are 1032 possible congurations of a
BMW 7 series car: A car with a given conguration comes
down the production line on average once every 4
months. The number of possible congurations continues
to increase, and BMW management has observed that the
prot potential in responding to this variability in demand
is substantial. For example, only a few cars per hundred
have head-up displays (in which key information is
projected onto the windshield), but these cars are
considerably more protable than standard models.
As delivery lead times decrease and product variability
increases, the area around the assembly line is not
sufcient to store the entire array of supplier parts at
BMWs assembly plants. Furthermore, reordering the parts
into the correct production sequence at the assembly plant
takes considerable space. Therefore, in the fall of 2003,
three major BMW auto assembly plants in Germany began
to require JIS delivery of parts such as engines.
Responding to this request required (a) a warehouse to
permit engines to be resorted prior to shipment along
with labor to do the sorting of engines, (b) a decoupling
point in the production process to allow standard engines
to be produced to stock and then nalized in the right
sequence, or (c) that the engine plant move to a
production schedule that matched the required sequence.
Although the idea of JIS production was attractive, there
was great concern that its implementation would result in
unacceptably low capacity utilization. Furthermore, the
300- to 600-km distance from the engine plant to
customer plants and the complexity of the product
implied that JIS implementation would be relatively
difcult. The question came down to where in the process
engines should be put into sequence, that is, at what point
in the process did need for exibility become greater than
need for efciency?
Although many companies attempt to achieve JIS
delivery without implementing JIS production through
combining postponement with a nal sorting operation in
a warehouse, experience shows that this solution becomes
unmanageable as companies look to strategic variability
to increase their margins. For example, as BMW exploits
new possibilities to customize engine and car options, it
will require an enormous warehouse to produce according
to a heijunka-dened schedule yet deliver JIS. A rethinking
of the balance between heijunka and JIS production is
clearly required.

4. Modeling the heijunka-JIS trade-off

In order to gain insight into the difference between
heijunka and JIS production we constructed a 10-workstation simulation model using Extend version 6

ARTICLE IN PRESS
504

ttmeir et al. / Int. J. Production Economics 118 (2009) 501507

A. Hu

(see www.imaginethatinc.com). All workstations have a

xed processing time of 50 s with the exception of
workstations 4 and 8, whose processing time is either
40 or 60 s depending on whether an engines work content
is low or high. Under heijunka, high and low work content
(difcult and easy) engines are alternated. Under JIS, the
sequence of difcult and easy engines is randomly
determined. In both cases, 50% of engines are modeled
as being difcult, 50% as being easy. Engines arrive to the
line at a rate of one engine every 50 s. There is no
processing time or interarrival time variability beyond the
difference in processing times for difcult and easy
engines at workstations 4 and 8. Each model run
represents 1 million seconds (approximately 4 weeks of
production time at the BMW engine plant), requiring
around 8 h on an average personal computer. The
structure of the model is shown in Fig. 1 below. All results
retrieved from this simulation are based on the specic
setup described above. They give insight into the behavior
of the system, but cannot be considered as generalizable.
Model results comparing the buildup of work in
process inventory (WIP) under JIS and heijunka sequencing are shown in Fig. 2a. The system has sufcient
capacity for the heijunka scenario, but becomes unstable
with the extra variability of JIS (i.e. WIP continually
increases over time).
One approach to regaining system stability under JIS is
to extend the cycle time by increasing the number of
seconds between engines. Increasing the interarrival rate
to 54 s completely eliminates the inventory buildup. This
represents a loss of capacity on the order of 7%, with
production during an 8-h shift decreasing from 576 to 533
engines. The choice between JIS and heijunka does not
have to be completely binary, however, and can instead be
seen as a continuum between these two extremes. The
actual sequencing of orders can be any compromise
between the required delivery sequence and a sequence
that makes the schedule as level as possible.
The choice between heijunka and JIS implies a cost
trade-off: as we have just observed, the responsiveness
and exibility of JIS comes at the price of a capacity buffer
(i.e. higher production costs), while pure heijunka requires
building responsiveness at the end of an inexible
production line (i.e. resorting costs). Therefore, we used
the simulation model to explore some intermediate points
on the heijunka-JIS continuum.
We began by considering how much inventory buildup
could be tolerated in the line, increasing the interarrival
time enough to stop the instability, but less than required
to eliminate WIP buildup. Fig. 2b shows the results with

cycle times of 50, 51 and 53 s. Moving from an interarrival

rate of 5051 s (which equals an increase of capacity of
2%) improves the situation already signicantly. The small
peaks can be compensated for by small buffer areas on the
assembly line to permit engines to wait without going out
of sequence, assuming that the wait was not too long.
Large engines go with large cars, so we also considered
the possibility that the engine sequence would already
have an element of smoothing completed by the auto
assembly plant which applies heijunka. This scenario
suggested a possibility to us of a sequence that would be
largely heijunka, but would allow a few engines to be out
of sequence. In other words, the production sequence
would follow heijunka discipline for the majority of
engines, but a small number of outliers would be
tolerated. We modeled this sequence as a series of 20
engines following heijunka rules (easy-difcult-easydifcult), followed by a break in the heijunka sequence
in the form of a second difcult engine, before restarting
the heijunka sequence. The interarrival rate was set to
51 s. The results, shown in Fig. 2c, suggest that partial
heijunka allows some reduction of the required capacity
buffer compared to the 53-s interarrival rate shown in Fig.
2b; in Fig. 2b an increase of interarrival time by 3 s still
leads to the formation of some WIP peaks whereas partial
heijunka as illustrated in Fig. 2c allows an increase of
interarrival time by only 1 s to fully eliminate WIP
buildup.
If BMW wants to take advantage of strategic variability
it has to eliminate re-work under both heijunka and JIS.
Fig. 3 illustrates how re-work disturbs the production and
how inventory starts to build up. In this model we assume
that an engine has a probability of 1% at every station to
get re-worked and 99% of going through. At an interarrival
rate of 50 s, the system is unstable and WIP becomes
innite under both heijunka and JIS because of the extra
time required for rework. Increasing the interarrival time
by two seconds (i.e., a capacity buffer of 4%) is enough to
stabilize the system and bring WIP under control; or in
other words, 1% rework requires 4% additional capacity to
avoid building up of WIP.
How do these insights compare to the situation at
BMWs engine plant? We began by collecting one year
(representing millions of records) of production data from
the SAP system to determine how many engines changed
place, and by how many positions. The production data
indicated that well over half of the engines changed
places, but the majority remained within 60 places of
their original position. A scant 15% of engines changed
more than 60 places. Only a very small percentage of

Fig. 1. Trading off between heijunka and JIS.

ARTICLE IN PRESS
ttmeir et al. / Int. J. Production Economics 118 (2009) 501507
A. Hu

505

80

WIP [engines]

70
60
50
JIS: 50 sec

40
30

Heijunka: 50 sec

20
10
0

100

200

300

400

500

600

800

700

Time [in thousands of seconds]

80
50 sec

WIP [engines]

70
60
50
40

51 sec

30
53 sec

20
10

WIP [engines]

450
400
350
300
250
200
150
100
50
0

100

200

400
600
300
500
Time [in thousands of seconds]

700

800

50 sec

51 sec

100

200

300
500
400
600
Time [in thousands of seconds]

700

800

Fig. 2. WIP for different scenarios of JIS, heijunka and partial heijunka: (a) JIS versus heijunka with a cycle time of 50 s, (b) JIS with cycle times of 50, 51
and 53 s, and (c) partial heijunka with cycle times of 50, 51 and 53 s.

engines changed place in the original sequence by more

than 200 places. Furthermore, it was relatively easy to
predict which engines would be delayed by many places:
Most were special engines that required substantial extra
testing procedures. The only engines that unexpectedly
faced a delay of several places were those that had quality
problems requiring rework.
Initial data analysis indicated that pure JIS would
require a capacity buffer on the order of 10% (Huttmeir,

2003). One could argue that such an investment in a

capacity buffer could be warranted given the expense
of the warehouse and sorting, and could even be
seen as a strategic investment as it would permit a
higher level of exibility and customer responsiveness.
Nevertheless, such thinking was not readily embraced
within the company due to a commitment to leanness and
the fact that utilization was a primary performance
measure.

ARTICLE IN PRESS
ttmeir et al. / Int. J. Production Economics 118 (2009) 501507
A. Hu

506

450
400

WIP [engines]

350
300

JIS: 50 sec

250
Heijunka: 50 sec

200
150
100

Heijunka: 52 sec

JIS: 52 sec

50
0

100

200

300
400
500
Time [in thousandsof seconds]

600

700

800

Fig. 3. The impact of rework on WIP for different scenarios of JIS and heijunka. Rework causes WIP to explode for both the JIS and heijunka scenarios.

We then considered the scenario in which the SAP/R3

system parameters were set to be less drastic in smoothing the production schedule, combined with a smaller
capacity buffer (on the order of 23%) and racks on the
line to collect and re-sort engines. The production data
suggested that this system would work well for the
approximately 75% of engines changing less than 60
places. The remaining 25%, however, would require
substantially larger buffers (which we estimated at
around 38 h of production).
Heijunka may naturally propagate in the BMW supply
chain in that large cars require large engines. A downstream producer following heijunka may well result in
substantial smoothing upstream. In such a case, the
approach that we have been discussing that is primarily
heijunka with space for regular exceptions might work
extremely well. For example, the sequence could be
planned so that an extra difcult engine could be added
once per time period (say, once per 20 min). This would
distribute the disruption but with more exibility than
pure heijunka, and could be achieved with a much smaller
capacity buffer.
Although the main concern within the BMW engine
plant concerning the move from heijunka to JIS was
with the reduction in capacity utilization, we observed
from the production data that the primary problem
came from rework. Under heijunka operating in combination with a postponement strategy, an engine is not
dedicated to a given customer until relatively late in its
production (referred to by BMW as late order assignment).
Under JIS, however, the dedication of the engine to
a customer is expected to occur relatively early in
production. Once the engine is dedicated, any changes
in sequence are much more disturbing. Success with
JIS will require almost complete elimination of rework.
Such sensitivity to rework generates an interesting
historical parallel to the original concept of JIT production,
in which removal of buffer inventories requires a radical

reduction in rework, production downtime, and material

shortages.
5. Summary and conclusions
Is it better for a manufacturing plant to use heijunka to
maximize its leanness, or to use JIS to maximize its
responsiveness? Our work at BMW indicates that the
answer may lie somewhere in the middle, with heijunka
used to smooth out the most extreme production values
with the remainder of production carried out JIS. It is
important to understand this trade-off, as it gives
essential insights into the bigger picture of trading off
between leanness and agility. We have blindly accepted
lean production practicesand leanness itselfas good
for competitiveness, but it is now time to reach a ner
understanding of these dynamics. Over time, we may nd
that the essential dynamic capabilities for competitive
operations will not come from congurations of routines
that drive leanness, but from congurations of routines
that permit essential exibility in spite of leanness. Also, it
may be that the combination of leanness and agility will
create new incentives to improve process reliability, given
the lack of tolerance for rework under JIS production.
References
Abdulmalek, F.A., Rajgopal, J., 2007. Analyzing the benets of lean
manufacturing and value stream mapping via simulation: a process
sector case study. International Journal of Production Economics 107,
223236.
Browning, T.R., Heath, R.D., 2008. Reconceptualizing the effects of lean
on production costs with evidence from the F-22 program. Journal of
Operations Management, doi:10.1016/j.jom.2008.03.009, 22pp.
Chase, R.L., 1993. Differing perceptions. The Total Quality Management
Magazine 5 (2), 1921.
Cusumano, M.A., 1994. The limits of lean. Sloan Management Review 1,
2732.
De Smet, R., Gelders, L., 1998. Using simulation to evaluate the
introduction of a kanban subsystem within an MRP-controlled

ARTICLE IN PRESS
ttmeir et al. / Int. J. Production Economics 118 (2009) 501507
A. Hu

manufacturing environment. International Journal of Production

Economics 5657, 111122.
de Treville, S., Antonakis, J., 2006. Can lean production be intrinsically
motivating? Contextual, congurational, and levels-of-analysis issues. Journal of Operations Management 24 (2), 99123.
Feare, T., 2003. Just-in-time, just-in-sequence. Modern Materials Handling.
Feitzinger, E., Lee, H.L., 1997. Mass customization at Hewlett Packardthe power of postponement. Harvard Business Review 75 (1),
116121.
Fujimoto, T., 1999. The Evolution of a Manufacturing System. Oxford
University Press, Toyota, NY.
Hall, R.W., 1983. Zero Inventories. Irwin, Homewood, IL.
Hines, P., Holweg, M., Rich, N., 2004. Learning to evolve: a review of
contemporary lean thinking. International Journal of Operations &
Production Management 24 (10), 9941011.
Hopp, W.J., Spearman, M.L., 1996. Factory Physics, rst ed. Irwin, Chicago.
Huttmeir, A., 2003. Competitive operations and trade-off analysis.
Master Thesis, University of Lausanne, Switzerland.
Kempfer, L.M., 2005. Speeding inline-vehicle sequencing. Material
Handling Management.
Lambert, D.M., Cooper, M.C., Pagh, J.D., 1998. Supply chain management:
implementation issues and research opportunities. The International
Journal of Logistics Management 9 (2), 120.
Liker, J.K., Wu, Y., 2000. Japanese automakers, US suppliers and supplychain superiority. Sloan Management Review, 8193.
McLachlin, R., 1997. Management initiatives and just-in-time manufacturing. Journal of Operations Management 15, 271292.
Mishina, K., Takeda, K., 1994. Toyota Manufacturing USA, Inc. Case 9-693019, Harvard Business School, Cambridge, MA, 22pp.
Monden, Y., 1983. Toyota Production System. Industrial Engineering
and Management Press, Institute of Industrial Engineers, Norcross,
GA.
Narasimhan, R., Swink, M.L., Kim, S.W., 2006. Disentangling leanness and
agility: an empirical investigation. Journal of Operations Management 24 (5), 440457.
Narasimhan, R., Talluri, S., Das, A., 2004. Exploring exibility and
execution competencies of manufacturing rms. Journal of Operations Management 22 (1), 91106.

507

Pietsch, D., 2002. Das neue Vertriebs- und Produktionssystem der BMW
Group-Kundenorientierter Vertriebs- und Produktionsproze (KOVP).
ZfAW (BMW), 4551.
Prenninger, J., 2001. Neue Anforderungen an eine Werksubergreifende
LogistikInnovationen im Bereich der Logistikkette Motor bei.
Gabler Verlag, BMW.
Prince, J., Kay, J., 2003. Combining lean and agile characteristics: creation
of virtual groups by enhanced production ow analysis. International
Journal of Production Economics 85, 305318.
Rinehart, J., 1997. After lean production: evolving employment practices in
the world auto industry. American Journal of Sociology, 12121214.
Sayer, A., 1986. New developments in manufacturing: the just-in-time
system. Capital & Class 10, 4371.
Schonberger, R.J., 1982. Japanese Manufacturing Techniques: Nine
Hidden Lessons in Simplicity. Free Press, New York.
Selto, F.H., Renner, C.J., Young, S.M., 1995. Assessing the organizational t
of a just-in-time manufacturing system: testing selection, interaction, and systems models of contingency theory. Accounting,
Organizations & Society 20 (7/8), 665684.
Shah, R., Ward, P.T., 2003. Lean manufacturing: context, practice bundles,
and performance. Journal of Operations Management 21, 129149.
Suri, R., 2003. QRM and POLCA: a winning combination for manufacturing enterprises in the 21st century. Technical Report, QRM Center,
University of Wisconsin, Madison, WI.
Suri, R., de Treville, S., 1986. Getting from just-in-case to just-in-time:
insights from a simple model. Journal of Operations Management 6
(3), 295304.
Teece, D.J., Pisano, G., Shuen, A., 1997. Dynamic capabilities and strategic
management. Strategic Management Journal 18 (7), 509533.
Trebilcock, B., 2006. Building a new supply chain. Modern Materials
Handling 61 (1), 32.
Van Hoek, R.I., 2001. The rediscovery of postponement: a literature
review and directions for research. Journal of Operations Management 19 (2), 161184.
Womack, J.P., Jones, D.T., Roos, D., 1990. The Machine that Changed the
World. Harper Collins Publishers, New York.
Yao, A.C., Carlson, J.G.H., 2003. Agility and mixed-model furniture
production. International Journal of Production Economics 8182,
95102.