The current issue and full text archive of this journal is available at

www.emeraldinsight.com/0960-0035.htm

Exploring the
Exploring the benefits of vendor benefits of VMI
managed inventory
Kazim Sari
Department of International Logistics and Transportation, 529
Beykent University, Istanbul, Turkey
Received March 2007
Revised May 2007
Abstract Accepted May 2007
Purpose The purpose of this paper is to explore the performance increase achieved by vendor
managed inventory (VMI) under different levels of outside supply capacity, demand uncertainty, and
lead time.
Design/methodology/approach The study uses discrete event simulation to explore the
performance increase achieved by VMI under different supply chain scenarios.
Findings The analysis suggests that when implementing a VMI program, the capacity restrictions
of suppliers have to be taken into consideration even though they have not participated in the
program. Furthermore, the results also show that unless the retailer provides additional information to
the distributor to resolve the uncertainty, higher levels of uncertainty in market demand create
significant reductions in the savings realized by VMI. Finally, this study proves that, regardless of
lead time horizons, VMI provides nearly the same level of performance increase as long as the ratio
of the retailers lead time to that of the suppliers remains constant.
Practical implications This analysis provides a means for practitioners to understand the impact
of various environmental and operational factors on the performance increase achieved by VMI so
they can better analyze their specific business conditions to prepare their organizations for more
successful VMI programs.
Originality/value Although there is a range of research focusing on VMI, only a few of those have
tried to identify the factors that play an important role in the failure of VMI programs. While
extending the current literature, this is the first study to explore the impact of outside supply capacity
on the performance improvements achieved by VMI.
Keywords Supply chain management, Vendor relations, Inventory management, Information strategy,
Simulation
Paper type Research paper

1. Introduction
Supply chain management (SCM) is:
. . . the efficient management of the end-to-end process, which starts with the design of the
product or service and ends with the time when it has been sold, consumed, and finally,
discarded by the consumer (Swaminathan and Tayur, 2003, p. 1387).
By coordinating different enterprises along the logistics network or establishing
business partnerships, SCM is concerned with finding the best strategy for the whole
supply chain (Simchi-Levi et al., 2003, p. 2). One important issue of finding the best
strategy for the whole supply chain is sharing product and production information International Journal of Physical
amongst supply chain members. It has been recognized that information sharing at Distribution & Logistics Management
Vol. 37 No. 7, 2007
the retailer level produces significant benefits for the supply chain by reducing the pp. 529-545
bullwhip effect (Lee et al., 1997a; Chen et al., 2000a, b; McCullen and Towill, 2002; q Emerald Group Publishing Limited
0960-0035
Dejonckheere et al., 2004; Ouyang, 2006; Li et al., 2006) and supply chain costs DOI 10.1108/09600030710776464
IJPDLM (Gavirneni et al., 1999; Gavirneni, 2002). However, in spite of these advantages,
37,7 retailers, most of the time, do not desire to engage in information sharing. This is due to
the fact that the primary beneficiary from information sharing is the manufacturers,
not the retailers (Lee et al., 2000; Yu et al., 2001, 2002; Zhao et al., 2002a, b).
Vendor managed inventory (VMI), also known as continuous replenishment or
supplier-managed inventory, is one of the most widely discussed partnering initiatives for
530 encouraging collaboration and information sharing among trading partners (Angulo et al.,
2004). Popularized in the late 1980s by Wal-Mart and Procter & Gamble (Waller et al.,
1999), it was subsequently implemented by many other leading companies from different
industries, such as Glaxosmithkline (Danese, 2004), Electrolux Italia (De Toni and Zamolo,
2005), Nestle and Tesco (Watson, 2005), Boeing and Alcoa (Micheau, 2005), etc. It is
a supply chain initiative where the vendor decides on the appropriate inventory levels of
each of the products and the appropriate inventory policies to maintain those levels.
The retailer provides the vendor with access to its real-time inventory level. In this
partnership program, the retailer may set certain service-level and/or shelf-space
requirements, which are then taken into consideration by the vendor. That is, in a VMI
system, the retailers role shifts from managing inventory to simply renting retailing space
(Simchi-Levi et al., 2003, p. 154; Mishra and Raghunathan, 2004).
VMI offers a competitive advantage for retailers because it results in higher product
availability and service level as well as lower inventory monitoring and ordering cost
(Waller et al., 1999; Achabal et al., 2000). For vendors, on the other hand, it results in
reduced bullwhip effect (Lee et al., 1997b; Disney and Towill, 2003a, b) and better
utilization of manufacturing capacity (Waller et al., 1999), as well as better synchronization
of replenishment planning (Waller et al., 1999; Cetinkaya and Lee, 2000).
While many benefits have been identified in the literature, there are also a number
of challenges that may exist in practice and that can potentially reduce the benefits
obtained from VMI or lead to failures in VMI programs. For instance, Spartan Stores, a
grocery chain, shut down its VMI effort about one year after due in part to VMI
vendors inability to deal with product promotions (Simchi-Levi et al., 2003, p. 161).
Similarly, Kmart cut a substantial amount of VMI contracts because they were not
satisfied with the forecasting ability of VMI vendors (Fiddis, 1997). Indeed, a few
research studies have been carried out to investigate the underlying reasons of these
failures (Aviv, 2002; Ovalle and Marquez, 2003; Angulo et al., 2004; Yao et al., 2007).
Most of these research studies are limited to identifying the factors that play an
important role in these failures. This may be because they only concentrate on the
internal dynamics of VMI. Indeed, the following two factors may be extracted from
these earlier studies as an explanation for failed VMI programs:
(1) sharing of outdated or inaccurate sales and inventory data due in part to a lack
of adequate information technology as well as a lack of mutual trust; and
(2) inaccurate demand forecast owing to the fact that under VMI, retailers are
excluded from the demand forecasting process.

Our research differs from those that have been carried out in previous studies in
one major aspect. Beside the internal dynamics of VMI, we have also considered the
performance of an outside supplier on the success of VMI programs. For example,
consider an outside supplier serving raw materials for VMI partners. The capabilities of
this supplier, which is not participated into VMI, may have significant impacts on the
benefits gained from VMI. For example, the poor performance of the non-participating Exploring the
supplier may lead to a failure in VMI. This factor, to the best of our knowledge, is not benefits of VMI
explored in supply chain literature. Exploring the impact of non-participant upstream
members on the performance of VMI is thus one major contribution of this research
study.
Moreover, when we investigate the relationship between the benefits of information
sharing and demand uncertainty, we see that the results of the previous research studies 531
differ considerably. Some show that the value of information sharing is higher at higher
levels of demand uncertainty (Lee et al., 2000; Zhang and Zhang, 2007), while others
demonstrate that information sharing is more beneficial at low levels of demand
uncertainty (Waller et al., 1999, Gavirneni, 2002; Lau et al., 2004). The reason behind
these differences may be differences in modelling assumptions, the types of supply chain
environment studied, or cost structures used. However, in all cases, these differences
among the contributions create confusion for SCM practitioners. In this study, we try to
explore the impact of demand uncertainty on the performance improvements achieved
by VMI. Providing a meaningful explanation of these differences in the literature is
another important contribution of this research study.
Unlike many prior analytical studies which have very restrictive assumptions for
the sake of mathematical tractability (Mishra and Raghunathan, 2004; Lee and Chu,
2005; Yao et al., 2007), we have used a simulation model in this study to investigate the
benefits of VMI under more realistic circumstances. The simulation approach has been
used extensively in the literature for analysing supply chain systems (Waller et al.,
1999; Zhao et al., 2002a, b; Angulo et al., 2004; Lau et al., 2004; Zhang and Zhang, 2007).
In this study, we considered a four-stage supply chain which consists of four echelons:
a manufacturing plant, a warehouse, a distributor, and a retailer. The plant has limited
manufacturing capacity and produces a single product. Each enterprise replenishes its
inventory from its immediate upstream enterprise.
The remainder of this study is organized as follows. The next section reviews the
previous literature related to information sharing and VMI with emphasis on the
benefits of VMI. Section 3 clarifies the methodology and development of the simulation
model. Setting of experimental design is identified in Section 4, followed by simulation
results. Conclusions are in the final section.

2. Related literature
There have been a number of analytical and simulation studies that have extensively
examined the factors affecting the benefits derived from information sharing (Gavirneni
et al., 1999; Lee et al., 2000; Yu et al., 2001, 2002; Gavirneni, 2002; Zhao et al., 2002a, b;
Simchi-Levi and Zhao, 2003; Lau et al., 2004; Croson and Donohue, 2005; Zhang and Zhang,
2007). Most of these research studies suggest that information sharing yields significant
performance improvements for the supply chain. Moreover, the improvements realized in
the performance of a supply chain enterprise through information sharing are also verified
by many empirical studies (Zhou and Benton, 2007).
Under certain conditions, however, the contribution of information sharing on
supply chain performance is substantially higher. Information sharing is more
beneficial under conditions where downstream level information is shared rather than
upstream level information (Lau et al., 2004; Croson and Donohue, 2005); and/or where
inventory policies are reformulated to make better use of shared information
IJPDLM (Gavirneni, 2002); and/or where higher levels of manufacturing capacity is available
37,7 (Gavirneni et al., 1999; Gavirneni, 2002; Simchi-Levi and Zhao, 2003; Lau et al., 2004);
and/or where supplier lead time is longer (Lee et al., 2000) and/or; where retailer lead
time is shorter (Moinzadeh, 2002). However, when we investigate the relationship
between the value of information and demand uncertainty, we see that the results of
the previous research studies differ considerably. Some show that value of information
532 sharing is higher at higher levels of demand uncertainty (Lee et al., 2000; Zhang and
Zhang, 2007), while others demonstrate that information sharing is more beneficial
under low levels of demand uncertainty (Waller et al., 1999; Gavirneni, 2002; Lau et al.,
2004).
On the other hand, we also realize that the benefits derived from information
sharing are not distributed equally between suppliers and retailers. For example,
analytical studies of Lee et al. (2000) and Yu et al. (2001, 2002) in a two stage supply
chain environment, with one supplier and one retailer, revealed that retailers benefits
are insignificant when compared to the benefits gained by suppliers. Similarly, Zhao
et al. (2002a, b) conducted a simulation study of a supply chain consisting of one
capacitated supplier and four retailers and found that retailers needed to be encouraged
to participate in information sharing, due to the fact that the benefits to retailers were
insignificant from information sharing.
These studies may explain why supply chain initiatives such as VMI have been
popularized in industry as a tool for encouraging retailers to participate in information
sharing. In order to better understand the underlying reasons behind the popularity of
VMI, we may also need to consider the research carried out by Lee and Chu (2005).
In their analytical study, Lee and Chu (2005) tried to identify the appropriate conditions
under which VMI is desirable for both parties. Their results indicate that VMI is
beneficial for both parties if the stock level desired by the supplier at the retailer
location is higher than that desired by the retailer. This means that suppliers with their
desired stock level at the retailer location can determine the retailers decision about
participating in VMI or not. At this point, the perceived higher cost of stockouts for
suppliers compared to retailers (Achabal et al., 2000), may explain the popularity of
VMI in different industries. Similarly, Yao et al. (2007) conducted an analytical study of
a two-stage supply chain to identify the appropriate conditions under which
implementing VMI is more desirable. Results from his analytical model indicate that
the cost savings realized from VMI are likely to be higher as more and more reduction
is achieved in ordering cost of retailers through VMI.
Other researchers, on the other hand, investigated the performance of VMI under
different conditions. Waller et al. (1999) carried out a simulation study in a two-stage
supply chain to analyze the impact of VMI under various levels of demand variability,
limited manufacturing capacity, and partial channel coordination. Their results
indicate that VMI provides greater inventory reductions and a higher utilization of
manufacturing capacity, mainly resulting from more frequent inventory reviews and
shorter intervals between deliveries.
Aviv (2002) constructed a model of two-level supply chain consisting of a retailer and
a supplier. In his model, Aviv (2002) concentrated on the abilities of the supplier and the
retailer to explain the uncertainty observed in customer demand. The analytical and
subsequent numerical study of Aviv (2002) indicates that the success of VMI programs
mainly depends on the relative forecasting skill of the supplier. That is, their results
indicate that the performance of the supply chain under VMI could be even worse than Exploring the
the traditionally managed supply chain unless the relative explanation power of the benefits of VMI
supplier is sufficiently good.
In a different research study, Ovalle and Marquez (2003) reached similar
conclusions as Aviv (2002). They carried out a simulation study by using system
dynamics modelling in order to evaluate the performance of various types of
collaboration strategies. They presented the simulation results of a four-stage supply 533
chain as described by Sterman (1989) in the article for the bear game. The results of
their study indicate that one important problem of VMI programs is the elimination
of retailers from inventory and forecasting decisions.
Finally, Angulo et al. (2004) explored the impact of sharing inaccurate and delayed
information on the performance improvements achieved by VMI in a four-stage supply
chain both for stationary and non-stationary demand structures. Statistical analysis of
the simulation outputs indicated that the performance of VMI substantially decreases
if the shared retail information is not up-to-date. However, they also showed that when
the inaccurate information is considered, unless it is highly inaccurate, does not create
substantial reductions in the performance of VMI.

3. The supply chain simulation model

At the initial stages of this research, we intended to use Microsoft Excel in constructing
the simulation model; however, research conducted by Keeling and Pavur (2004)
indicated that it might be possible for errors to occur in the random numbers generated
by Microsoft Excel. Therefore, in order to eliminate this potential problem, we have
used Crystal Ball, an Excel add-in published by Decisioneering. It is a popular risk
analysis and forecasting program that uses Monte Carlo simulation in a spreadsheet
environment.
Two supply chain structures are considered in the model for comparison purposes.
The first structure is a supply chain operated under traditional ways of doing business
(TSS) and the second structure is a supply chain model operated under a VMI program.
In both systems, an (R, S) inventory control policy is used for replenishment decisions.
Here, R indicates the review interval and S indicates the order-up-to level. Review
interval (R) is chosen as one week. Order-up-to level, however, is updated at the
beginning of each week to reflect changes in demand patterns. The simplified
simulation logic and the flowchart are represented in Figure 1.
Under TSS, all supply chain members strive to develop local strategies for
optimizing their own organizations without considering the impact of these strategies
on the performance of other supply chain members. Moreover, since there is no
information sharing between members, upstream stages are unaware of actual demand
information at the market place. Here, while creating demand forecasts and inventory
plans, supply chain members use only replenishment orders placed by their immediate
downstream member. Moreover, each supply chain member uses an installation stock
concept (i.e. members consider only their own inventory level) in their
production/inventory decisions. The sequence of events followed by a supply chain
member under TSS is outlined as follows:
.
The member receives the delivery from its immediate upstream member, which
was ordered L periods ago (the lead-time is L periods). If the member is the plant,
L is the production lead time.
IJPDLM
37,7 START Parameters of Parameters of
Demand Structure Capacity (CAP)

534 DEMAND
GENERATION
DETERMINATION OF THE
PLANT'S CAPACITY

Replenishment Point of Sales

Orders Data

DEMAND FORECAST SUPPLY CHAIN DEMAND FORECAST

TSS VMI (The Retailer and The
(Exponential Smoothing) TYPE
Distribitor)

DETERMINE ORDER CALCULATE

CALCULATE AMOUNT ORDER-UP-TO LEVEL
ORDER-UP-TO LEVEL (The Retailer and The (The Retailer and The
Distribitor) Distribitor)

Inventory Level
of Retailer
DEMAND FORECAST
Replenishment (The Warehouse and The
Orders Plant)

DETERMINE ORDER
AMOUNT DETERMINE ORDER
AMOUNT
(The Warehouse and The
Distribitor)

SATISFY INCOMING
ORDER

REPORT

Average Supply Total Supply Customer

Chain Inventory Chain Cost Service Level

Figure 1.
END
Flow chart of the
simulation model
 .
The member observes the order placed by its immediate downstream member. Exploring the
If the member is the retailer, the order is the market demand. benefits of VMI
.
The member fulfils the customer orders (plus backorders if there are any)
by on-hand inventory, and any unfulfilled customer orders are backordered.
The member analyzes the historical replenishment orders placed by its
immediate downstream member for forecasting. Based on this demand forecast,
the member updates its order-up-to point. If the member is the retailer, historical 535
market demand data is analyzed. The order-up-to point of the member at stage k,
S k ; estimated from the observed demand is as follows (Nahmias, 1997, p. 278):
 
bk
Sk F 21
k 1
bk hk

where F k : is the distribution function of the demand realized by the member

at stage k. Similarly, bk and hk are back-order and holding costs of the member at
stage k, respectively. Here, parameters of the demand distribution, F k :; are
updated at the beginning of each week by using the exponential smoothing
method (Nahmias, 1997, p. 74) to reflect changes in demand patterns.
.
The member decides how many units to order from its immediate upstream
member. The quantity of the order is equal to the difference between the
order-up-to level and inventory position. If the member is the plant, a production
order is placed. Here, the plant, because of its limited manufacturing capacity,
cannot always produce enough to bring its inventory position up to the updated
value of S. In these cases, the plant makes full capacity production by
backordering the remaining requirement. This modification of order-up-to policy
for the case of limited production capacity provides an optimal solution for
uncertain demands (Gavirneni et al., 1999; Federgruen and Zipkin, 1986a, b).

Under VMI, on the other hand, the retailer provides the distributor with access to its
real-time inventory level as well as its point-of-sales (pos) data. In return, the
distributor takes the responsibility of managing the inventories of the retailer. That is,
under VMI, the distributor does not only need to take its own inventories into account
while making inventory plans, but also the inventories of the retailer. Therefore, under
this structure, the distributor uses the echelon stock concept (Clark and Scarf, 1960;
Axsater and Rosling, 1993) in replenishment planning. All other echelons of the supply
chain (the plant and the warehouse), on the other hand, operated in the same way as in
TSS. In order to compute the echelon order-up-to levels of the retailer and the
distributor, the heuristic developed by Shang and Song (2003) is used. Again, in this
supply chain, the exponential smoothing method is used to update the order-up-to level
at each week.
Although the literature suggests that VMI may result in significant reductions in
retailers lead times (Waller et al., 1999), we do not include this reduction in our
simulation model. Instead, we assume that the retailers lead times remain constant
after VMI implementation. Indeed, here we have concentrated on information sharing
and centralized decision-making aspects of VMI only. Therefore, we may surely state
that the results obtained from this study constitute a lower bound for the benefits
obtained from VMI.
IJPDLM The cost structures for the supply chain members in the simulation model are
37,7 assumed to be as follows; the unit back-order costs per week for the plant, the
warehouse, the distributor, and the retailer are $5, 11, 18, and 25, respectively. The unit
inventory costs per week for the plant, the warehouse, the distributor, and the retailer
are $0.25, 0.50, 0.75, and 1.00, respectively.

536 3.1 Retailers demand structure

Although normal distribution is more widely used in supply chain research studies, the
g distribution is used here to represent the customer demand realized by the retailer. This
is due to the fact that there are some limitations in the normal distribution in representing
demand structures. For example, normal distribution allows the occurrence of negative
customer demand. Therefore, in order to avoid this unrealistic situation, some restrictive
assumptions have to be included in the model (Waller et al., 1999; Zhao et al. 2002a, b; Lau
et al., 2004). The g distribution, on the other hand, does not have such problems because it
allows only non-negative values. Moreover, the g distribution is flexible in that it can
represent a wide variety of demand structures. Keaton (1995), for instance, states that
choosing g distribution is an effective choice to represent the demand patterns.
There are two parameters of the g distribution. These are shape a and scale b
parameters. The mean and the variance of the distribution can be expressed as ab and
ab 2 ;, respectively. In the simulation model, we assume that the shape parameter of the
demand distribution is 15 a 15: The scale parameter b; on the other hand, is
assumed to be a stochastic variable in the form of equation (1).
 
2p
bt 20 season sin t 2
52
In equation (2), bt is the scale parameter of the g distribution in week t.
The variability in the scale parameter of the demand distribution allows us to generate
both seasonal and non-seasonal customer demands. For example, while assigning zero
to the season constant produces non-seasonal demand pattern, assigning non-zero
values results in seasonality in customer demand. A representative histogram of
the market demand for the selected parameters is generated in Figure 2 to clarify the
distribution of the market demand to the readers.
Three demand structures representing different combinations of seasonality are
used in this study. These are customer demand with no seasonality (SDV), customer
demand with medium level of seasonality (MDV), and customer demand with high
level of seasonality (HDV). The values of the season constant for each demand
structure are determined as 0, 2, and, 4, respectively. The values of the season constant

0.25
0.2
Probability

0.15
0.1

Figure 2. 0.05
Histogram of the market 0
demand when a 15 and
117

157

197

237

277

317

357

397

437

477

517

557

597

bt 20
Market Demand
are selected in such a way that both non-seasonal and seasonal customer demands with Exploring the
different strengths are generated. For example, while SDV represent the non-seasonal
customer demand, MDV and HDV represent the demand structures with seasonal
benefits of VMI
swings of the size of approximately 10 and 20 percent of average demand, respectively.

3.2 Verification and validation of the simulation

In order to verify that the simulation program performs as intended, the conceptual 537
model is divided into three parts: demand generation and determination of total
manufacturing capacity, forecasting and production/inventory planning, and order
fulfillment and reporting. Each part is designed separately so that more efficient and
effective debugging is made possible. Moreover, the combined simulation model is also
traced and tested with the results calculated manually.
Later, in order to validate the simulation output, the random demand variables
generated in the simulation model are plotted on a scatter diagram. Then, it is
validated that the intended demand structure is generated. The supply chain model
above was simulated for 1,128 weeks. The initial parameters of the forecasting model
were estimated with the first 400 weeks of simulation run, which were removed later
from the output analysis to eliminate the worm-up period effect. Therefore, the rest of
the data was used for effective simulation output analysis. In order to reduce the
impact of random variations, the same random numbers were used to simulate both
systems. That is, same customer demand was generated for both types of supply chain
system. In addition to this variance reduction technique, 15 replications for each
combination of the independent variables were conducted.

4. Experimental design
Four independent factors are considered in the experimental design. These are; type
of supply chain (SCTYPE), available production capacity of plant (CAP), uncertainty
in customer demand (DV), and replenishment lead times (L). The number of levels of
these variables and their values are listed in Table I.
Factor SCTYPE refers to the way the supply chain is operated. Specifically, this
factor indicates whether the supply chain is operated under TSS or VMI. The factor
CAP is expressed as the ratio of the plants total capacity to the total market demand.
Total capacity of the plant is distributed to each week, equally. The factor L denotes
the replenishment lead times between each member of the supply chain. Finally, the
factor DV indicates the level of uncertainty seen in market demand.
Three kinds of factors are used as dependent variables in the experimental design in
order to evaluate benefits gained from VMI. These factors are average inventory level in
the supply chain (INV), total cost for the entire supply chain (TSC) and customer service
level of the retailer (CSL). TSC is the sum of the inventory holding costs of all members in

Levels
Independent factors 1 2 3 4 5

SCTYPE TSS VMI

DV SDV MDV HDV Table I.
L 1 4 7 Independent factors of
CAP 1.10 1.20 1.30 1.40 1.50 the experimental design
IJPDLM the supply chain and backorder cost of the retailer. Here, we include the back order cost
37,7 of the retailer only, because all other back order costs are internal costs within the entire
supply chain and they are not actually incurred. Factor CSL is the percentage of
customer demand satisfied by the retailer through the available inventory.

538 5. Simulation output analysis

The output from the simulation experiments is analyzed using the MANOVA
procedure of the SPSS. The MANOVA procedure is more appropriate for our model
because it considers the correlation between the dependent variables in the
experimental design. For more detailed information about MANOVA see the paper
by Hair et al. (1998, p. 331). Selected MANOVA results are presented in Table II.
MANOVA results in Table II show that at 5 percent significance level, SCTYPE has
significant impacts on all three-performance factors, which indicates that VMI has
substantial influences on the performance of the supply chain. The performance of
each type of supply chain is presented in Table III.

Dependent variables
CSL INV TSC (a)
Source F-value Pr . F F-value Pr . F F-value Pr . F

SCTYPE 85.379 0.0000 3,818.611 0.0000 127.424 0.0000

CAP 26.536 0.0000 1,371.075 0.0000 3.455 0.0000
L 444.785 0.0000 39,476.629 0.0000 4,090.439 0.0000
DV 1,130.268 0.0000 6.572 0.0000 570.928 0.0000
SCTYPE *CAP 10.854 0.0000 21.995 0.0000 4.727 0.0000
SCTYPE *L 28.735 0.0000 248.518 0.0000 1.148 0.3180
CAP *L 7.782 0.0000 157.211 0.0000 5.366 0.0001
SCTYPE *CAP *L 0.245 0.9820 18.197 0.0000 1.123 0.2821
SCTYPE *DV 20.900 0.0000 33.560 0.0000 40.252 0.0000
CAP *DV 8.321 0.0000 58.013 0.0000 4.413 0.0000
SCTYPE *CAP *DV 0.604 0.7746 0.110 0.9989 0.846 0.5622
L *DV 85.479 0.0000 47.666 0.0000 72.943 0.0000
SCTYPE *L *DV 2.218 0.0287 2.325 0.0988 1.223 0.2821
CAP *L *DV 3.293 0.0009 4.107 0.0000 5.294 0.0000
SCTYPE *CAP *L *DV 0.316 0.9953 1.021 0.3450 0.260 0.9985
Table II.
Selected MANOVA Note: aBased on residual analysis, log transformation of TSC was made to satisfy the assumptions of
results ANOVA

95 percent confidence interval

Performance measures SCTYPE Average Lower bound Upper bound

CSL (percentage) TSS 94.52 94.35 94.70

VMI 96.08 95.95 96.21
TSC ($) TSS 701,761 685,510 718,013
Table III. VMI 577,899 547,198 608,189
Performance of each type INV (unit) TSS 1,956 1,949 1,963
of supply chain VMI 1694 1686 1701
Examination of Table III reveals that compared to TSS, the reduction in average Exploring the
inventory level in the supply chain is 13.4 percent. Similarly, the reduction in total benefits of VMI
supply chain cost varies from 6.5 to 43.3 percent with an average around 17.6 percent.
The results also indicate that VMI significantly increases the retailers service level.
For example, while the customer service level under TSS is 94.3 percent on the average,
it is around 96 per cent under VMI. Therefore, these results lead us to conclude that
VMI produces substantial increases in the retailers customer service level while 539
decreasing average inventory level and the total cost of the entire supply chain.
Actually, these findings are simple and intuitively expected for us, so we will not
concentrate on them further. Instead, we will concentrate on how a non-participating
members manufacturing capacity (CAP) affects the benefits of VMI as well as the
uncertainty in customer demand (DV) and lead times (L). For this purpose,
performance of VMI under various capacity levels (CAP), demand uncertainty (DV),
and lead times (L) are produced in Figure 3.

5.1 Capacity of non-participated member

MANOVA results in Table II show that at 5 percent significance level, the interaction
effect between CAP and SCTYPE has significant impacts on all three dependent
factors. This means that capacity level of non-participated member has a significant
influence on the performance of VMI for all three performance measures.
Examination of Figure 3 shows that the reduction amount in total supply chain cost
is significantly higher at larger levels of capacity at the plant. For example, the
reduction amount in total supply chain cost increases from 14.6 to 21 percent, on
the average when capacity ratio (CAP) increases from 1.10 to 1.50. In addition, when we
consider the average inventory level of the supply chain we see a very similar situation
with supply chain cost. Finally, when the customer service level is considered, we see
that under all levels of manufacturing capacity, VMI produces higher fill rates than
TSS. Therefore, these findings show us that VMI produces substantially higher
benefits at higher levels of the manufacturing capacity. This result is very interesting.

24 98
24 98
TSC and INV (%)
TSC and INV (%)

20 97
20 97
Reduction in
Reduction in

CSL (%)

CSL (%)

16 96
16 96
12 95
12 95
8 94
8 94
4 93
4 93
0 92
0 92
1 4 7 SDV MDV HDV
L DV

25 97
TSC and INV(%)

20 96
Reduction in

CSL (%)

15 95 TSC
INV
10 94 CSL of TSS
5 93 CSL of VMI
Figure 3.
0 92 Overall performance of
1.10 1.20 1.30 1.40 1.50
VMI
CAP
IJPDLM That is, in this supply chain, the plant does not participate in the VMI program, but its
37,7 manufacturing capacity has a significant influence on the performance of VMI.
The result obtained here is consistent with previous research studies (Gavirneni
et al., 1999; Lee et al., 2000; Gavirneni, 2002; Simchi-Levi and Zhao, 2003; Lau et al.,
2004). That is, in these earlier studies, researchers suggest that benefits obtained from
information sharing substantially decrease with respect to in-house capacity
540 restriction. Our research, however, extended the results obtained from these earlier
studies by indicating that capacity restriction of non-participated suppliers also
substantially reduces the benefits derived from VMI.

5.2 Uncertainty in customer demand

MANOVA results in Table II show that at 5 percent significance level, the interaction
effect between DV and SCTYPE has a significant impact on all performance factors.
This indicates that uncertainty in customer demand has a significant influence on the
performance of VMI for all three performance measures.
Examination of Figure 3 shows that the reduction in total supply chain cost and the
average inventory level achieved by the VMI program reduce as the level of demand
uncertainty increases. For example, we see from Figure 3 that when there is no
seasonality in customer demand, the reduction in total supply chain cost and average
inventory level are realized as 24 and 16 percent, respectively. However, when there is a
high-degree of seasonality, these savings decreased to 15 percent for the total supply
chain cost and 9.3 percent for the average inventory level. Moreover, when we consider
the other performance criteria, we see that under all levels of demand uncertainty, VMI
produces higher level of service.
The results obtained here show us that the uncertainty in customer demand
substantially reduces the benefits obtained from VMI, as also indicated in some earlier
studies (Waller et al., 1999; Gavirneni, 2002; Lau et al., 2004). In contrast with these results,
the findings of other researchers (Lee et al., 2000; Zhang and Zhang, 2007) show that
information sharing is more beneficial at higher levels of demand uncertainty.
The underlying reasons for this apparent contradiction between these studies may
be differences in modelling assumptions, the supply chain environment studied, or
cost structures used. However, among these reasons the most influential ones are related to
the type of information sharing strategy implemented. While some types of information
sharing strategies involve almost all relevant descriptions about uncertainty of probable
future demand that exists in the system, some others convey only partial information to
help supply chain partners in resolving uncertainty. Namely, answers of the following two
questions play a critical role in determining the value of information sharing under
different levels of demand uncertainty. These are:
(1) Which types of information are shared among supply chain partners?
(2) Under what conditions are these information shared?

Here, let us consider two information sharing strategies. In the first strategy,
information of a retailer about an upcoming promotion is fully transferred to the
upstream members along with sales and inventory data. In the second strategy,
however, only inventory and sales data are shared. Of course, the benefits obtainable
from these two strategies under different levels of demand uncertainty vary. That is,
the benefits obtained from the former strategy, under the market conditions where
much of the uncertainty stems from product promotions, are substantially higher. On Exploring the
the contrary, the value of the second strategy substantially decreases as uncertainty benefits of VMI
comes from product promotions increases.
As a result, it can be said that information sharing is more beneficial at higher levels
of demand uncertainty as long as the type of information shared can help supply chain
members in resolving the uncertainty observed in customer demand. Otherwise, higher
levels of demand uncertainty will definitely reduce the benefits obtained from 541
information sharing as has also been seen in this research.

5.3 Lead times

MANOVA results in Table II show that at 5 percent significance level, the interaction
effect between L and SCTYPE has a significant impact on the customer service level
and supply chain inventory, but it does not have any significant impact on the total
supply chain cost. This means that while reduction in average inventory level and
increase in customer service level realized from VMI change with lead times, reduction
of supply chain cost gained from VMI does not change with lead times.
Examination of Figure 3 shows that the inventory reduction obtained from VMI is
higher at shorter lead times, but by very small amounts. For example, while the
inventory reduction is 13.90 percent at a lead time of one week, it is 13.10 percent at a
longer lead time of seven weeks. Similarly, we also see that VMI provides a little higher
performance in fill rates at shorter lead times. That is, while it produces 1.95-points
higher fill rate than TSS when the lead times are one week, it provides 1.14-points
higher fill rate when the lead times are seven weeks. However, these slight performance
improvements in service level and average inventory level are not reflected in total
supply chain cost as can be seen in Table II and Figure 3. Therefore, from these results,
we can conclude that regardless of the lead time horizon, VMI produces nearly the
same level of performance increase.
Current literature on information sharing indicates that performance increase
achieved by information sharing under longer supplier lead time is quite substantial
and it increases with respect to the supplier lead time (Lee et al., 2000). Similarly,
the literature also suggests that performance increase achieved by information sharing
is lower under longer retailer lead time (Moinzadeh, 2002). This paper extends the
current literature and provides valuable information for SCM practitioners by
indicating that VMI or information sharing provides nearly the same level of
performance increase in supply chains where lead times are short or long, as long as
the ratio of the retailers lead time to that of the suppliers remains constant.

6. Conclusions
This paper explores the benefits of VMI in supply chains involving four stages by
providing important insights into the performance of VMI under various supply chain
scenarios. One major contribution of this research is to explore the influence of capacity
restriction of an outside supplier on the performance of VMI. Moreover, the
performance increase achieved by VMI under various supply chain scenarios, which
are characterized by demand uncertainty and lead times, are also explored extensively.
Through comprehensive simulation experiment and subsequent statistical analysis of
the simulation outputs, we make the following important findings:
IJPDLM (1) We examine that there are substantial decreases in the performance increase
37,7 obtained from VMI as the uncertainty in customer demand increases, unless the
shared information provides additional information to the vendor to resolve
uncertainty in customer demand.
(2) We examine that the performance of outside supplier plays an important role
for the success of VMI programs. Our simulation study indicates that as the
542 manufacturing capacity of the outside supplier decreases, the benefits obtained
from VMI also decreases substantially. This finding, interestingly, shows that
success of VMI programs does not depend only on internal dynamics of VMI,
but also on external factors. In addition, we also realize that capacity restriction
of a non-participated supplier is more crucial when the demand uncertainty
in customer demand is high. This is because the higher demand uncertainty in
customer demand also reduces the benefits derived from VMI.
(3) Finally, we examine the impact of lead times on the performance of VMI.
We observe that, regardless of lead time horizons, VMI provides nearly the
same level of performance increase as long as the ratio of the retailers lead time
to that of the suppliers remains constant.
In conclusion, managers in a supply chain enterprise have to make careful benefit/cost
analysis while making their decisions regarding VMI implementation under conditions
where the performance of upstream members is poor and/or uncertainty in customer
demand is high. Although this study provides some insights into VMI and its
relationship with outside suppliers as well as demand uncertainty, there are some
limitations that we have to state. First, in the simulation model, we assumed that the
retailers lead time remains constant after VMI implementation; however, VMI may, in
fact, significantly reduce the retailers lead time. Second, we considered a serial supply
chain structure with one member at each echelon. This is only a simplified case and in
future researches, modelling more realistic supply chain structures will better explain
the results obtained from this research. Third, we assume that the members in the
supply chain use modified or simple order-up to policy to make their inventory decisions;
however, there are other inventory or production policies that can be included in the
model. Forth, the cost structure used in the model only represents one special case.

References
Achabal, D.D., McIntyre, S.H., Smith, S.A. and Kalyanam, K. (2000), A decision support system
for vendor managed inventory, Journal of Retailing, Vol. 76 No. 4, pp. 430-54.
Angulo, A., Nachtmann, H. and Waller, M.A. (2004), Supply chain information sharing in a
vendor managed inventory partnership, Journal of Business Logistics, Vol. 25 No. 1,
pp. 101-20.
Aviv, Y. (2002), Gaining benefits from joint forecasting and replenishment process: the case of
auto-correlated demand, Manufacturing & Service Operations Management, Vol. 4 No. 1,
pp. 55-74.
Axsater, S. and Rosling, K. (1993), Notes: installation vs echelon stock policies for multilevel
inventory control, Management Science, Vol. 39 No. 10, pp. 1274-80.
Cetinkaya, S. and Lee, C.Y. (2000), Stock replenishment and shipment for vendor-managed
inventory systems, Management Science, Vol. 46 No. 2, pp. 217-32.
Chen, F., Drezner, Z., Ryan, J.K. and Simchi-Levi, D. (2000a), Quantifying the bullwhip effect in a Exploring the
simple supply chain: the impact of forecasting, lead times and information, Management
Science, Vol. 46 No. 3, pp. 436-43. benefits of VMI
Chen, F., Drezner, Z., Ryan, J.K. and Simchi-Levi, D. (2000b), The impact of exponential
smoothing forecasts on the bullwhip effect, Naval Research Logistics, Vol. 47 No. 4,
pp. 269-86.
Clark, A.J. and Scarf, H. (1960), Optimal policies for a multi-echelon inventory problem, 543
Management Science, Vol. 6 No. 4, pp. 475-90.
Croson, R. and Donohue, K. (2005), Upstream versus downstream information and its impact on
the bullwhip effect, System Dynamics Review, Vol. 21 No. 3, pp. 249-60.
Danese, P. (2004), Beyond vendor managed inventory: the Glaxosmithkline case, Supply Chain
Forum: An International Journal, Vol. 5 No. 2, pp. 32-9.
De Toni, A.F. and Zamolo, E. (2005), From a traditional replenishment system to
vendor-managed inventory: a case study from the household electrical appliances
sector, International Journal of Production Economics, Vol. 96 No. 1, pp. 63-79.
Dejonckheere, J., Disney, S.M., Lambrecht, M.R. and Towill, D.R. (2004), The impact of
information enrichment on the bullwhip effect in supply chains: a control engineering
perspective, European Journal of Operational Research, Vol. 153 No. 3, pp. 727-50.
Disney, S.M. and Towill, D.R. (2003a), Vendor-managed inventory (VMI) and bullwhip
reduction in a two level supply chain, International Journal of Operations & Production
Management, Vol. 23 No. 6, pp. 625-51.
Disney, S.M. and Towill, D.R. (2003b), The effect of vendor managed inventory dynamics on the
bullwhip effect in supply chains, International Journal of Production Economics, Vol. 85
No. 2, pp. 199-215.
Federgruen, A. and Zipkin, P. (1986a), An inventory model with limited production capacity and
uncertain demands I: the average-cost criterion, Mathematics of Operations Research,
Vol. 11 No. 2, pp. 193-207.
Federgruen, A. and Zipkin, P. (1986b), An inventory model with limited production capacity and
uncertain demands II: the discounted-cost criterion, Mathematics of Operations Research,
Vol. 11 No. 2, pp. 208-15.
Fiddis, C. (1997), Manufacturer-retailer Relationships in the Food and Drink Industry: Strategies
and Tactics in the Battle for Power, Financial Times Retail & Customer Publications,
London.
Gavirneni, S. (2002), Information flows in capacitated supply chains with fixed ordering costs,
Management Science, Vol. 48 No. 5, pp. 644-51.
Gavirneni, S., Kapuscinski, R. and Tayur, S. (1999), Value of information in capacitated supply
chains, Management Science, Vol. 45 No. 1, pp. 16-24.
Hair, J., Anderson, R.E., Tatham, R.L. and Black, W.C. (1998), Multivariate Data Analysis, 5th ed.,
Prentice-Hall, Englewood Cliffs, NJ.
Keaton, M. (1995), Inventory control under gamma demand and stochastic lead time, Journal of
Business Logistics, Vol. 16, pp. 107-31.
Keeling, K.B. and Pavur, R.C. (2004), Numerical accuracy issues in using excel for simulation
studies, Proceedings of the 2004 Winter Simulation Conference, pp. 1513-8.
Lau, J.S.K., Huang, G.Q. and Mak, K.L. (2004), Impact of information sharing on inventory
replenishment in divergent supply chains, International Journal of Production Research,
Vol. 42 No. 5, pp. 919-41.
IJPDLM Lee, C.C. and Chu, W.H.J. (2005), Who should control inventory in a supply chain, European
Journal of Operational Research, Vol. 164 No. 1, pp. 158-72.
37,7 Lee, H., Padmanabhan, V. and Whang, S. (1997a), Information distortion in a supply chain:
the bullwhip effect, Management Science, Vol. 43 No. 4, pp. 546-58.
Lee, H., Padmanabhan, V. and Whang, S. (1997b), The bullwhip effect in supply chains, Sloan
Management Review, Vol. 38 No. 3, pp. 93-102.
544 Lee, H., So, K.C. and Tang, C.S. (2000), The value of information sharing in a two-level supply
chain, Management Science, Vol. 46 No. 5, pp. 626-64.
Li, J., Sikora, R., Shaw, M.J. and Tan, G.W. (2006), A strategic analysis of inter organizational
information sharing, Decision Support Systems, Vol. 42 No. 1, pp. 251-66.
McCullen, P. and Towill, D.R. (2002), Diagnosis and reduction of bullwhip in supply chains,
Supply Chain Management: An International Journal, Vol. 7 No. 3, pp. 164-79.
Micheau, V.A. (2005), How Boeing and Alcoa implemented a successful vendor managed
inventory program, Journal of Business Forecasting, Vol. 24 No. 1, pp. 17-19.
Mishra, B.K. and Raghunathan, S. (2004), Retailer- vs vendor-managed inventory and brand
competition, Management Science, Vol. 50 No. 4, pp. 445-57.
Moinzadeh, K. (2002), A multi-echelon inventory system with information exchange,
Management Science, Vol. 48 No. 3, pp. 414-26.
Nahmias, S. (1997), Production and Operations Analysis, Irwin/McGraw-Hill, Homewood, IL.
Ovalle, O.R. and Marquez, A.C. (2003), The effectiveness of using e-collaboration tools in the
supply chain: an assessment study with system dynamics, Journal of Purchasing &
Supply Management, Vol. 9 No. 4, pp. 151-63.
Ouyang, Y. (2006), The effect of information sharing on supply chain stability and the bullwhip
effect, European Journal of Operational Research, Vol. 182 No. 3, pp. 1107-21.
Shang, K.H. and Song, J.S. (2003), Newsvendor bounds and heuristic for optimal policies in serial
supply chains, Management Science, Vol. 49 No. 5, pp. 618-38.
Simchi-Levi, D. and Zhao, Y. (2003), The value of information sharing in a two stage supply
chain with production capacity constraints, Naval Research Logistics, Vol. 50 No. 8,
pp. 888-916.
Simchi-Levi, D., Kaminsky, P. and Simchi-Levi, E. (2003), Designing and Managing the Supply
Chain, 2nd ed., McGraw-Hill, Homewood, IL.
Sterman, J. (1989), Modeling managerial behavior: misperceptions of feedback in a dynamic
decision making experiment, Management Science, Vol. 35 No. 3, pp. 321-39.
Swaminathan, J.M. and Tayur, S.R. (2003), Models for supply chain in e-business, Management
Science, Vol. 49 No. 10, pp. 1387-406.
Waller, M.A., Johnson, M.E. and Davis, T. (1999), Vendor-managed inventory in the retail
supply chain, Journal of Business Logistics, Vol. 20 No. 1, pp. 183-203.
Watson, E. (2005), Nestle switches to vendor managed inventory with Tesco, Food
Manufacture, Vol. 80 No. 8, p. 20.
Yao, Y., Evers, P.T. and Dresner, M.E. (2007), Supply chain integration in vendor-managed
inventory, Decision Support Systems, Vol. 43 No. 2, pp. 663-74.
Yu, Z., Yan, H. and Cheng, T.C.E. (2001), Benefits of information sharing with supply chain
partnerships, Industrial Management & Data Systems, Vol. 101 No. 3, pp. 114-9.
Yu, Z., Yan, H. and Cheng, T.C.E. (2002), Modelling the benefits of information sharing-based
partnerships in a two-level supply chain, Journal of Operational Research Society, Vol. 53
No. 4, pp. 436-46.
Zhang, C. and Zhang, C. (2007), Design and simulation of demand information sharing in a Exploring the
supply chain, Simulation Modelling Practice and Theory, Vol. 15 No. 1, pp. 32-46.
Zhao, X., Xie, J. and Zhang, W.J. (2002a), The impact of information sharing and ordering
benefits of VMI
co-ordination on supply chain performance, Supply Chain Management: An International
Journal, Vol. 7 No. 1, pp. 24-40.
Zhao, X., Xie, J. and Leung, J. (2002b), The impact of forecasting model selection on the value of
information sharing in a supply chain, European Journal of Operational Research, Vol. 142 545
No. 2, pp. 321-44.
Zhou, H. and Benton, W.C. Jr (2007), Supply chain practice and information sharing,
Journal of Operations Management.

Further reading
Mason-Jones, R. and Towill, D.R. (2000), Coping with uncertainty: reducing bullwhip behaviour
in global supply chains, Supply Chain Forum: An International Journal, Vol. 1 No. 1,
pp. 40-5.

Corresponding author
Kazim Sari can be contacted at: kazims@beykent.edu.tr

To purchase reprints of this article please e-mail: reprints@emeraldinsight.com

Or visit our web site for further details: www.emeraldinsight.com/reprints
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.