7th IFAC Conference on Manufacturing Modelling, Management,

and Control
International Federation of Automatic Control
June 19-21, 2013. Saint Petersburg, Russia

Impact of Backorder on Supply Chain Performance – an Experimental Study

V. Madhusudanan Pillai* and T. Chinna Pamulety**

Mechanical Engineering Department, National Institute of Technology Calicut,
Kerala, India,
*(e-mail: [email protected])
** (e-mail:[email protected])

Abstract: Increase in order variance from downstream stage to upstream stage in a supply chain is called
bullwhip effect. Bullwhip effect is one of the performance measures of a supply chain. In the present
study the impact of backorder in supply chain performance is studied under small lead time of one period
by conducting experiments under lost sales and backorder settings using supply chain role play software
package. The performance of supply chain under both settings is compared using Wilcoxon statistical
test. The results show that the variance of orders is high under backorder than the lost sales setting. The
study concludes that backorder is also one of the causes of bullwhip effect.
Keywords: Bullwhip effect, Supply chain, Inventory, Performance measures, Statistical test, Backorder.

causes of bullwhip effect reported in the literature are
1. INTRODUCTION
categorized into operational causes and behavioural causes
Supply chain consists of various stages that are directly or (Paik and Bagchi 2007, Bhattacharya and Bandyopadhyay
indirectly involved in fulfilling the customer requirements 2011). Operational causes include Demand forecast
(Chopra et al. 2010). The performance of a supply chain can updating, Order batching, Variation in prices/Sales
be measured in-terms of bullwhip effect, supply chain fill promotions/Discounts, Rationing and shortage gaming,
rate, total cost of the supply chain and total inventory held Replenishment strategy, Number of levels in supply chain,
(Bhattacharya and Bandyopadhyay 2011). Lack of communication and coordination among the stages in
a supply chain, Lack of information transparency, Long lead
Increase in demand or order variability when we move from time, Supply variability, Capacity limit, Variability in
downstream stage to upstream stage in a supply chain is customer demand, Multiplier effect, Lack of synchronization,
called bullwhip effect (Lee et al. 1997). The term “bullwhip” Local optimization without global vision and Improper
was originally coined by the logistics executives at the control system. Literature shows that the behavioural causes
Procter & Gamble (P&G). This is one of the serious issues in of the bullwhip effect are identified by conducting
supply chain management. It has been observed in the experiments using beer distribution game. Various
commercial operations of many industries and a few of them behavioural causes include misperception of feedback
includes Hewlett-Packard (HP) (Lee et al. 1997), Procter & (Sterman 1989), lack of training and/or learning (Wu and
Gamble (P&G) (Lee et al. 1997), Machine tool industry Katok 2006), overreaction to incoming order or fear of empty
(Anderson et al. 2000), Clothing supply chain (Disney and stock (Croson and Donohue 2003) and behaviour of under-
Towill 2003), Grocery retailer in United Kingdom (Ge et al. estimating the value of information (Nienhaus et al. 2006).
2004), Phillip electronics (Kok et al. 2005), Semiconductor
equipment industry (Terwiesch et al. 2005), Campbell soup Most of the experimental studies in literature (Sterman 1989,
(Odonnell et al. 2006) and Wal-Mart (Bhattacharya and Croson and Donohue 2003, Croson and Donohue 2005, 2006,
Bandyopadhyay 2011). Bullwhip can affect the supply chain Wu and Katok 2006, Cantor and Katok 2012) are conducted
in many ways. It leads to excessive inventory or poor under backorder settings. But, backorder creates variation in
customer service level, revenue loss, inactive transportation, orders as the replenishment quantity varies considerably in
misguided capacity plans, missed production plans (Lee et al. backorder cases. This variation in replenishment quantity
1997), market share loss (Wright and Yuan 2008), addition of may get reflected while placing orders. Hence, we can say
unnecessary costs in the form of stock-out or inventory or that backorder is also one of the contributing factors to
obsolescence cost (Shukla et al. 2009), negative impact on bullwhip effect. Croson and Donohue (2002) reported that the
the performance of a supply chain (Nienhaus et al. 2006). experimental work is an important complement to theoretical
Reduction in bullwhip effect (BWE) can improve the work. Tokar (2010) highlighted the need for behavioural
performance and increase the profit of a supply chain research in logistics and supply chain management in which
(Caloiero et al. 2008). human objects are used for conducting the experiments. The
aim of this paper is to test the impact of backorder on
Since the presence of bullwhip degrades the performance of a bullwhip effect (supply chain performance) under a small
supply chain, many researchers tried to identify the various lead time of one period by experimentation. The experiments
causes of it and methods to reduce it in their studies. Various are conducted in backorder and lost sales settings and the

978-3-902823-35-9/2013 © IFAC 1932 10.3182/20130619-3-RU-3018.00477

2013 IFAC MIM
June 19-21, 2013. Saint Petersburg, Russia

results are analysed using Wilcoxon statistical test. The the form of a bar chart. The downstream inventory
performance of the supply chain role play software package information sharing reduced the bullwhip effect significantly
which is used for conducting the experiments in the present than the upstream inventory information sharing.
study is also validated by comparing the results available in
Sterman 1989. Lack of customer demand distribution information was one
of the reasons for the presence of bullwhip effect in Sterman
2. LITERATURE REVIEW (1989) study. Croson and Donohue (2006) analyzed the
performance of a supply chain under known and stationary
The performance of a supply chain can be evaluated by customer demand distribution to all players. Theory suggests
experimentation or simulation or analytical methods. Beer that bullwhip does not exist under known and stationary
distribution game is used to conduct experiments. The beer customer demand distribution but the experimental results
distribution game is a role play game which simulates a had showed the presence of bullwhip. The reason for its
supply chain with four stages viz. retailer, wholesaler, presence was the cognitive limitations of the participants in
distributor and factory. Participants take the role of a stage that study. They also studied the effect of sharing inventory
and decide (by intuition) how much to order from their information to all stage members as a bar chart in order to
suppliers based on their current stock situation and customer analyze the impact of inventory information on the
orders. All stages have a common goal of minimizing costs performance of a supply chain. It is found that the inventory
while avoiding out-of-stock situations. The cost includes information sharing has reduced the bullwhip effect
inventory holding cost and shortage cost. A brief summary on significantly when compared to first study which considered
the experimental studies reported in the literature are given sharing of known and stationary customer demand
below. distribution. Wu and Katok (2006) studied the impact of
training and communication on the bullwhip effect and
Sterman (1989) used beer distribution game for analyzing the reported that it reduced the bullwhip effect. Cantor and Katok
performance of a four-stage serial supply chain (2012) examined the production smoothing in serial supply
experimentally under a pattern of customer demand (the chains experimentally. The decrease in standard deviation of
demand level is changing from one level to another level in orders as we moves up the supply chain is called production
one step) which was not known to the participants of the smoothing. They conducted four types of experiments by
game. The four stages or players involved in his supply chain simplifying the four-stage serial supply chain of beer
experiment are retailer, wholesaler, distributor and factory. distribution game to two-stage serial supply chain. First one
Each player receives the orders from their downstream stage is conducted under known stationary customer demand
and tries to meet the orders, and then takes decision about the distribution of UNIF(0,8). Second one is conducted under
order quantity to be placed with upstream stage, customer demand distribution with predictable seasonal
independently without consulting the other players. The component. The customer demand in the second experiment
players reported that the reason for a large variance in their follows a cyclic demand of zero in odd time periods and eight
orders is due to their inability to predict the pattern of in even time periods. These experiments are conducted under
customer demand. Impact of Point of Sale (PoS) data on the short lead time, i.e., two periods between retailer and factory,
performance of a supply chain was studied by Croson and and one period of production delay at factory. The results
Donohue (2003). They conducted two sets of experiments. show that the retailer and factory do not amplify the orders
First set was conducted by sharing customer demand under both experimental conditions. The third and fourth
distribution and second one was conducted by sharing PoS experiments are conducted in a similar way as in the first and
data along with the customer demand distribution. It is second experiments respectively, but there is a cost for
observed that the impact of PoS data on the performance of change of order size (per unit) in current period when
supply chain was not clear when it is shared along with compared to previous period order size. Results show that the
customer demand distribution. Steckel et al. (2004) studied smoothing orders is more pronounced under costly order
the impact of lead time (time taken by the order to reach from change in seasonal demand setting when compared to
a downstream stage to its next upstream stage and the time costless order change in seasonal demand setting. There is no
taken for the shipment quantity to reach the downstream statistical evidence of order smoothing in costly order change
stage from its immediate upstream stage) and PoS data on in stationary demand distribution setting when compared to
bullwhip effect and found that smaller the lead time, the costless order change in stationary demand distribution
lesser is the bullwhip effect. This study also showed that the setting.
benefits of the PoS data sharing depend on the pattern of
customer demand. Croson and Donohue (2005) conducted Most of the above experimental studies were conducted
experiments to know the impact of sharing of upstream and under backorder environment and for a lead time of two or
downstream inventory information on bullwhip effect. They more periods. In the present competitive market, customers
conducted three experiments by sharing no information, may not be willing to wait for the products because of its
upstream inventory information and downstream inventory availability at the competitors. So, investigation of the
information. In the upstream inventory information sharing performance of supply chains under lost sales business
experiment, each stage is able to view the inventory position environment is relevant in the present market conditions. If
of its upstream stages and the inventory position of the time between order and delivery is large, it can act as a
downstream stages is shared in the downstream inventory major reason for confusion at the mind of decision maker
information sharing experiment. The information is shared in while placing order. Order placed at one point, generally, is

1933
2013 IFAC MIM
June 19-21, 2013. Saint Petersburg, Russia

available after lead time and it meets the demand of some and total supply chain inventory are also measured along with
distant future period/s. So, lead time can contribute variance of orders to evaluate the performance of supply
considerably to variation in orders (Metters 1997). Hence, the chain under both settings and the results are given in next
experiments in the present study are conducted with a small Section.
lead time of one period under backorder and lost sales
business environments. 4. RESULTS
The performance measures available from the supply chain
3. DETAILS OF EXPERIMENTAL STUDIES
role play game are consolidated and they are shown in
CONDUCTED
Figures 1 to 5. The average value of each performance
Experiments are conducted on a single product four-stage measure over six supply chains in each setting is given in
serial supply chain under lost sales and backorder settings, Table 1. Figures 1 and 2 show the variance of orders placed
for testing the impact of backorder on the performance of a by each stage and Figures 3 and 4 show the total inventory at
supply chain. The supply chain role play game software each stage under both settings. The total supply chain
package developed is used for conducting the experiments inventory, for the backorder and lost sales settings, is shown
under both settings. The unique feature of this software in Figure 5.
package is that it facilitates to set and evaluate the
performance of the supply chain under different market
environment such as lost sales and backorder. We can set the
supply chain under different information sharing settings in
this role play game. It also provides flexibility in setting the
supply chain parameters for each stage such as lead time,
initial inventory, holding cost and shortage cost. There was a
technical session before conducting the experiments in which
the bullwhip effect phenomenon is explained. There was a
practice session for 10 periods, before the actual play, which
helped the participants to understand the meaning and
interpretation of the terminology used and displayed by the
supply chain role play game software package. It also helped
to understand the features and the working of the software Fig. 1. Variance of orders placed by each stage under lost
package developed. sales environment.
In this study, 48 candidates participated and each one
manages one stage in the four-stage supply chain and thus
formed 12 identical supply chains. This means that there are
six supply chains under lost sales setting and remaining six
are under backorder. The participants are students from under
graduate, post graduate and Master of Business
Administration (MBA) programmes. Most of them are from
Industrial Engineering and Management specialization. The
duration of the experiment was not revealed to the
participants and it was conducted for 55 periods. Each
participant place orders to the upstream stage at each period
for replenishment and manages his or her own inventory. The
size of the order is decided by intuition with the objective of
maximizing the filling of downstream stage orders and Fig. 2. Variance of orders placed by each stage under
minimizing the inventory held. The order placed by stage i at backorder environment.
the end of period t reaches the stage (i + 1) in the beginning
of period (t + 1) and the quantity shipped by stage (i + 1) in
period (t + 1) reaches stage i in the beginning of period (t +
2). It is also assumed that weekly customer demand follows
normal distribution with mean 20 units and standard
deviation 5 units. Chan and Chan (2010) suggest that normal
distribution is a good choice for customer demand. Since,
order placed at the end of first period received at the
beginning of third period, an initial inventory of 40 units is
assigned to each stage. The data from period 7 to 48 are
considered for performance analysis as in Steckel et al.
(2004). Generally, this type of experiment is conducted to
analyse the bullwhip effect using variance of orders placed by Fig. 3. Total inventory of each stage under lost sales
each stage. In this study, total inventory held at each stage environment.

1934
2013 IFAC MIM
June 19-21, 2013. Saint Petersburg, Russia

magnitude of amplification between stages in the present

study is same as in the Sterman’s (1989) study but the values
are different. This difference may be due to the level of
customer demand and lead time considered. In the present
study, customer demand distribution considered is normal
distribution N (20, 5) whereas in Sterman’s study step-up
demand (initially the demand was at 4 units, later it was
raised to 8 units) was considered. The lead time considered in
the present study is one period at all stages whereas in
Sterman (1989), it was four periods for first three stages and
three periods for last stage. A similar type of validation can
be seen in Croson and Donohue (2006). They used similar
supply chain with information sharing but demand was
Fig.4. Total inventory of each stage under backorder following uniform distribution (Unif (0,8)) and lead time was
environment. same as in Sterman (1989).

Table.2 Comparison of magnitude of amplification

Magnitude of amplification
Sl. 2
2
3
2
4
2
Supply chain
No. 2 2 2
1 2 3
1 Sterman 1989 1.77 1.96 1.60
2 Present study 2.42 3.37 1.83

6. STATISTICAL TESTS
Sign and Wilcoxon statistical tests are conducted to draw
Fig.5. Total supply chain inventory. conclusions on the impact of backorder in the performance of
serial supply chain. Sign test is used to know the presence of
Table 1. Average values of each performance measure bullwhip effect in the supply chain under backorder and lost
sales settings. Wilcoxon test is used to know the impact of
Supply Stage name backorder on the variance of orders, total end period
chain Retailer Wholesaler Distributor Factory inventory at all stages and total supply chain inventory. All
setting the statistical tests are evaluated at 5% significance level with
Variance of orders one-sided p-values. Results are tabulated, and which are
Lost sales 28.25 51.56 90.61 137.60 found to be significant are highlighted in the respective
Backorder 45.68 110.44 372.02 689.62 tables.
Total inventory
Lost sales 217.33 513.17 596.5 1260.33 6.1 Sign test
Backorder 420.67 1020.83 2778.83 2261.83
Total supply chain inventory Since the variable under consideration (variance of orders)
Lost sales 2587.33 has continuous distribution, sign test (see Siegel and
Backorder 6482.17 Castellan 1988) is used to test the presence of bullwhip effect
(Croson and Donohue 2006). Here, for each supply chain, an
increase in the variance of orders placed between consecutive
stages is coded as a success and a decrease as a failure. If
5. VALIDATION OF EXPERIMENTAL RESULTS
variance of orders between consecutive stages is same, it is
coded as zero and it is dropped from the analysis. The
The above experimental results are validated by comparing probability of success or failure is 0.5. Success is represented
with the existing results in the literature under same business by a plus (+) sign and a failure with minus (–) sign. The
environment. Sterman (1989) conducted experiments under sample size (n) is the sum of the plus and the minus signs. If
backorder business environment for a single product in a X represents the number of plus signs, then the probability of
four-stage serial supply chain. The magnitude of getting X or more plus signs is calculated using the Binomial
amplification between the stages from their experimental distribution. If this probability is less than the significance
results is compared with the present study experimental level fixed, the hypothesis must be rejected. The hypotheses
results under same business environment. The magnitude of for the present study are given below.
amplification between two stages is equal to the ratio of
2
variance of orders ( σ ) of stage (i+1) to stage i. The results Hypothesis-H1: Bullwhip effect does not occur under
are given in Table 2. From Table 2, the pattern of change in backorder with a lead time of one period.

1935
2013 IFAC MIM
June 19-21, 2013. Saint Petersburg, Russia

Hypothesis-H2: Bullwhip effect does not occur under lost the stages and end period inventory of all stages. The
sales with a lead time of one period. hypotheses framed are given below. The number of
observation in x is equal to 24 (l = 24) because there are six
The above procedure is followed to test the presence of supply chains under backorder setting and each supply chain
bullwhip effect under backorder and lost sales using variance has four stages. Similarly, group y contains the appropriate
of orders and the details of the test results are given in Table performance measure considered for all stages in a supply
3. chain under lost sales setting and y has 24 observations (m =
24). The hypotheses framed are tested by following the above
Table 3. Result of sign test for different supply chain settings procedure. The test results are tabulated in Table 4.
Sl. Success rate Sign test parameters Hypothesis-H3: Variance of orders in the stages of the supply
Hypothesis
No. (%) n X p chain under backorder and lost sales are equal.
1 H1:Backorder 88.88 18 16 0.0006
2 H2:Lost sales 83.33 18 15 0.0037 Hypothesis-H4: The magnitude of end period inventory in the
stages of the supply chain under backorder and lost sales are
From Table 3, it is found that the p-value of both settings is equal.
less than the significance level fixed and hence the success
rate is higher than the chance rate of 50%. Thus, the Since the p-value of this test is less than the significance level
hypotheses framed should be rejected. Hence, the study fixed, the hypotheses framed under these settings must be
concludes that the bullwhip effect occurs in supply chains rejected. Hence, it is inferred that the variance of orders and
under backorder and lost sales under a lead time of one the total end period inventory in the stages of the supply
period. chain is significantly high under backorder than lost sales.

6.2 Wilcoxon-Mann-Whitney test Table 4. Test for impact of backorder on magnitude of

variance of orders and total end period inventory in the stages
This test is usually termed as Wilcoxon test. Wilcoxon test of the supply chain
(see Siegel and Castellan 1988) is conducted to know the
Sl. Wilcoxon Parameters
impact of backorder on the magnitude of variance of orders Hypothesis
No. Wx Wy z p
(Croson and Donohue 2006), total inventory at each stage
Magnitude of variance of orders
and total supply chain inventory. This test procedure is as
follows: Let l be the number of observations in a group x and H3: Backorder Vs
1 712 464 2.567 0.0051
m be the number of observations in a group y. The groups x Lost sales
and y are the variance of orders placed by all stages in a Magnitude of end period inventory in the stages
supply chain under two different settings such as backorder H4: Backorder Vs
2 711 465 2.546 0.0054
and lost sales. The observations of both the groups are Lost sales
combined and assigned ranks starting from one to (l + m) by
arranging them in ascending order. If observations are equal, Another hypothesis framed with respect to total supply chain
average rank is assigned to both and the statistics Wx, Wy, inventory is given below. Since the number of supply chains
and z (if l or m greater than 10) are calculated. Wx and Wy under each setting is less than 10, this hypothesis is tested by
are the sum of the rank of observations which belongs to using Wilcoxon test when sample size of l or m is less than
groups x and y respectively, and 10. For the present problem, l = 6 and m = 6, and the
W x  0 .5   W
statistical test is conducted at 5% significance level. The test
z  x
(1) results are given in Table 5.
 Wx
where, Hypothesis-H5: Total supply chain inventory under
l ( K  1) 2 lm ( K  1) backorder and lost sales are same.
W x  , Wx  and K lm
2 12
The hypothesis framed needs to be rejected when the Table 5. Test for impact of backorder on total supply chain
probability value for the calculated z value is less than the inventory
fixed significance level α. (If l or m less than 10, the
probability associated with the occurrence under null Sl. Wilcoxon Parameters
Hypothesis
hypothesis of any Wx as extreme as the observed value is No. Wx Wy p
determined (Siegel and Castellan 1988) and the null H5:Backorder Vs
1 52 26 P(Wx ≥ 52)=0.0206
hypothesis must be rejected, if this probability is less than the Lost sales
fixed significance level α.)
Since P(Wx ≥ 52) = 0.0206 ( From Table J given in Siegel
For the present problem, it is assumed that group x contains and Castellan, 1988) which is less than the significance level
the appropriate performance measure considered for all fixed, the hypothesis framed must be rejected. Hence, the test
stages in a supply chain under backorder setting. Two confirmed that the total supply chain inventory under
hypotheses are framed to study the effect of backorder in the backorder is higher than the lost sales setting.
performance measures such as variance of orders placed by

1936
2013 IFAC MIM
June 19-21, 2013. Saint Petersburg, Russia

7. CONCLUSIONS effect in supply chains. International Journal of

Production Economics, 85, 199-215.
Based on the experimental studies conducted on a four-stage Ge, Y., Yang, J.B., Proudlove, N. and Spring, M. (2004).
serial supply chain, it is found that the bullwhip exists in the System dynamics modelling for supply-chain
supply chain irrespective of the business environment. management: A case study on supermarket chain in the
However, it is noticed that the magnitude of variance of UK. International Transactions in Operational
orders placed (bullwhip effect) under backorder is higher Research, 11, 495-509.
than the lost sales case. That is, it is possible to say that Kok, T.D., Janssen, F., Doremalen, J.V., Wachen, E.V.,
backorder environment promotes the bullwhip effect. The Clerkx, M. and Peeters, W. (2005). Philips electronics
magnitude of variance of orders, end period inventory of all synchronizes its supply chain to end the bullwhip effect.
stages and total supply chain inventory under backorder is Interfaces, 35 (1), 37-48.
significantly higher than the lost sales. That is, the backorder Lee, H.L, Padmanabhan, V. and Whang, S. (1997). The
has significant impact on the performance of the supply bullwhip effect in supply chains. Sloan Management
chain. This study is conducted by keeping the lead time at a Review, 38 (3), 93-102.
practically minimum level of one period. In today’s Metters, R. (1997). Quantifying the bullwhip effect in supply
competitive market, lost sales, rather than backorder may be chains. Journal of Operations Management, 15, 89-100.
the business environment faced by a supply chain. This study Nienhaus, J., Ziegenbein, A. and Schoensleben, P. (2006).
reveals that lost sales has much better performance compared How human behavioural amplifies the bullwhip effect. A
to backorder business environment. study based on the beer distribution game online.
In this paper, it has been shown by experiment that backorder Production planning and control, 17(6), 547-557.
promotes bullwhip effect. This may also has to be established Odonnell, T., Maguire, L., Mcivor, R. and Humphreys, P.
through simulation study. (2006). Minimizing the bullwhip effect in a supply chain
using genetic algorithms. International Journal of
REFERENCES Production Research, 44(8), 1523 – 1543.
Paik, S. K. and Bagchi, P. K. (2007). Understanding the
Anderson, E.D., Fine, C.H. and Parker, G.G. (2000). causes of the bullwhip effect in a supply chain.
Upstream volatility in the supply chain: The machine International Journal of Retail & Distribution
tool industry as a case study. Production and operations Management, 35 (4), 308 – 324.
Management Society, 9, 239-261. Shukla, V., Naim, M.M. and Yaseen, E.A. (2009). Bullwhip
Bhattacharya, R. and Bandyopadhyay, S. (2011). A review of and backlash in supply chains. International Journal of
the causes of bullwhip effect in a supply chain. Production Research, 47(23), 6477-6497.
International Journal of Advanced Manufacturing Siegel, S. and Castellan, N.J. (1988). Nonparametric
Technology, 54, 1245 – 1261. statistics for the behavioural sciences. McGraw Hill
Cantor, D.E. and Katok, E. (2012). Production smoothing in international editions, Singapore.
a serial supply chain: a laboratory investigation. Steckel, J.H., Gupta, S. and Banerji, A. (2004). Supply chain
Transportation Research Part E, 48, 781 – 794. decision making: will shorter cycle times and shared
Caloiero, G., Strozzi, F. and Comenges, J.Z. (2008). A supply point-of-sale information necessarily help? Management
chain as a series of filters or amplifiers of the bullwhip Science, 50 (4), 458-464.
effect. International Journal of Production Economics, Sterman, J.D. (1989). Modeling managerial behaviour:
114, 631-645. Misperceptions of feedback in a dynamic decision
Chan, H.K. and Chan F.T.S. (2010). A review of making experiment. Management Science, 35(3), 321-
coordination studies in the context of supply chain 339.
dynamics. International Journal of Production Terwiesch, C., Ren, Z.J., Ho, T.H. and Cohen, M.A. (2005).
Research, 48, 2793-2819. An empirical analysis of forecast sharing in the
Chopra, S., Meindl, P. and Kalra, D.V. (2010). Supply chain semiconductor equipment supply chain. Management
management: Strategy, planning and operation. Pearson Science, 51 (2), 208-220.
Prentice Hall, Noida. Tokar, T. (2010). Behavioural research in logistics and
Croson, R. and Donohue, K. (2002). Experimental economics supply chain management. The International Journal of
and supply-chain management. Informs, 32(5), 74-82. Logistics Management, 21(1), 89-103.
Croson, R. and Donohue, K. (2003). Impact of POS data Wright, D. and Yuan, X. (2008). Mitigating the bullwhip
sharing on supply chain management: an experiment. effect by ordering policies and forecasting methods.
Production and Operations Management, 12(1), 1-12. International Journal of Production Economics, 113,
Croson, R. and Donohue, K. (2005). Upstream versus 587-597.
downstream information and its impact on the bullwhip Wu, D.Y., Katok, E. (2006). Learning, communication and
effect. System Dynamics Review, 21(3), 249 – 260. bullwhip effect. Journal of Operations Management, 24,
Croson, R. and Donohue, K. (2006). Behavioural causes of 839-850.
the bullwhip effect and the observed value of inventory
information. Management Science, 52 (3), 323-336.
Disney, S.M. and Towill, D.R. (2003). The effect of vendor
managed inventory (VMI) dynamics on the bullwhip

1937