Description
The bullwhip effect is among the most studied effects in supply chain management research. Part of that research explores behavioural causes using the experimental context of the beer distribution game. This paper explicitly relates personality characteristics to performance in supply chain tasks within this experimental context.

403
THE BEER GAME REVISITED: RELATING RISK-TAKING
BEHAVIOUR AND BULLWHIP EFFECT

Gwenny Ruël, Dirk Pieter van Donk and Taco van der Vaart

University of Groningen, Faculty of Management and Organisation, Groningen, The Netherlands,
{g.c.ruel,d.p.van.donk,j.t.van.der.vaart }@rug.nl

ABSTRACT
The bullwhip effect is among the most studied effects in supply chain management research. Part of that
research explores behavioural causes using the experimental context of the beer distribution game. This
paper explicitly relates personality characteristics to performance in supply chain tasks within this
experimental context. This paper shows that differences in personality characteristics such as risk taking,
efficacy, ambiguity, and locus of control lead to differences in performance. Low risk taking persons have
on average higher back order costs, and lower inventory costs, while high risk taking persons show the
opposite cost structure. Further analysis suggests differences with respect to this personality
characteristic, which results in differences in decision making behaviour. Further research using more
controlled experiments and larger samples will further test these results that might have important
consequences for the role of humans in managing real-life supply chains.

Keywords: beergame, personality characteristics, human behaviour, bullwhip effect, supply chain
management, experiments

INTRODUCTION
Supply chain management strives for an optimal functioning of a supply chain. Generally, optimal is seen
as satisfying demand of ultimate consumers while minimising inventories at different positions in the
chain, keeping order backlogs low (or even not having them at all) and minimising costs. In a real-life
organisational context managing a supply chain is difficult due to uncertainties in demand of consumers
and the dependency of performance of one chain member upon the decisions and actions of other supply
chain members. Moreover, the feedback on the effect of decisions (such as ordering and shipping) usually
is delayed and/or indirect. All in all, this might lead to the well-known and regularly investigated
bullwhip effect. The bullwhip effect can be described as the phenomenon where the variance in demand
for wholesaler, distributor and manufacturer is larger than for the retailer and amplifies upstream (e.g. Lee
et al., 1997). Lee et al. (1997) show this effect in real-life settings (which is replicated by others) and
address four operational causes of the problem: errors in demand signal processing, inventory rationing,
order batching, and price variations. There is a still growing amount of research that further explores,
mostly using an analytical approach, the bullwhip effect and comes up with further analysis of the
problem and ways to solve it.
Another line of research focuses on behavioural causes of the bullwhip effect. The optimal
functioning of a supply chain seems to be often distorted by specific behaviour of individual decision
makers in the chain. One of the distortions is the overestimation of demand, resulting in the bullwhip
effect. The bullwhip effect can be illustrated using the beer game (Beer Distribution Game) originally
developed and used at MIT (Sterman, 1989). Jacobs (2000) developed an internet version based on the
original manual version of the game that used boards and cards. The internet version (and the original
version) is ideal for conducting experiments in a supply chain context. Croson and Donohue (2002)
review a number of the experiments done so far. The majority of papers up to this point have manipulated
the information and instructions given to players in or before the game. Although most of them
acknowledge the influence of human behaviour on the game results, human behaviour itself is often
treated as a black-box. The present study tries to explore some of the factors underlying this human

404
behaviour As such this paper follows one of the suggestions for future work of Croson and Donohue
(2002) to explore the relationship between individual characteristics (such as patience, risk-neutrality and
abstract thinking) and performance of supply chains tasks (p. 82). Nauta et al. (2002) found that there are
significant differences in personality between planners, production managers and marketing/sales
employees, suggesting that this might be a relevant and fruitful direction. The aim of this paper is to
explore whether differences in personality characteristics can explain task performance in a supply chain
context. From the personality characteristics our main experimental variable will be the “risk-taking” of a
(team of) player(s). Our main goal is to investigate to what extent this characteristic is related to both the
decision making strategy and to the performance of the chain as a whole.
The paper is structured as follows. The next section explores and introduces a number of personality
characteristics and psychological concepts. The second section will describe the methodology, data
collection and the experimental context of the beer game. The third section will describe the main results
that will be further explored in the subsequent section. Here, we also attempt to better understand the
differences in decision making of different persons. The fifth section discusses the results. Final remarks
and future research issues are presented in the conclusions.

THEORETICAL BACKGROUND
Nienhuis et al. (2002) characterized the strategy of players by means of two categories: players using a
“safe harbour” or a “panic” strategy. The “safe harbour” strategy involves holding on to average/standard
deviation; that is the player orders the amount that was in average ordered from him during the last
periods, plus an amount to create safety stock, depending on the standard deviation of orders he received.
Players with a “panic” strategy want to keep level of stock. Therefore each order they receive, they pass
on to the supplier. Therefore we were interested in the question to what extent the personality
characteristic “risk-taking” of the (team of) players is related to both the decision making strategy and to
the performance of the chain as a whole. As this study aims to explore the influence of personality
characteristics, we also wanted to look at other personality characteristics that could be relevant in this
situation. As previous research suggests, one of the main characteristics of the beer game is that persons
feel to be in a complex, difficult to judge situation. As such the situation can be qualified as ambiguous.
According to Barrick and Mount (1991) emotional stability refers to an individual’s level of self
confidence and balance with respect to work, and to the individual’s response to new and ambiguous
tasks. Therefore we wanted to investigate to what extent “tolerance for ambiguity”, “self efficacy”, and
“locus of control” influence the decision making and performance. Self efficacy can be described as the
belief in one’s ability to perform a task (Bandura, 1977) and is considered to be an important predictor of
performance (Bandura, 2003). Locus of control (Rotter, 1990) is known as the tendency of people to
attribute the causes of their behaviour to either themselves (internal locus of control) or to environmental
factors (external locus of control). This means that people with an internal locus of control believe that
they can control outcomes, whereas people with an external locus of control have the feeling that their
performance is the product of circumstances beyond their control.

METHODOLOGY

Participants and procedure
In a pilot study among PhD-students, we tested a first version of a questionnaire on risk-taking and other
psychological traits. After analysing the data the questionnaire was cut back from 42 items to 33 items.
Then we asked first year undergraduate students to complete the (reduced) questionnaire and to
participate in a game that was to be played in a four member team, on the computer. About 120 students
filled in the questionnaire and eventually 56 of them participated in the Beer Distribution Game. They
were divided over 14 homogeneous groups: 5 low risk taking groups, 3 middle risk taking groups and 6
high risk taking groups. After screening the data, we concluded that two groups (or particular members in
that group) were outliers and showed very deviating behaviour that could significantly influence the
overall picture. That left us with 4 low risk, 3 middle risk and 5 high risk groups.
In the second step we created three groups: low, middle and high risk-taking. Out of each group we
selected teams of four players that were homogeneous on our main experimental variable, the amount of
risk taking behaviour. The game was played using a static demand and doubling the demand in the fifth
period, as in the original Beergame experiment described in Sterman (1989). No other changes were made

405
during the 40 rounds played and participants were not allowed to talk to each other. All relevant decisions
during the game were recorded and participants were asked to answer a number of questions relating to
their strategy, impressions and experiences.

Measurements
First of all we conducted reliability analyses on the scales of items relating to personality traits. The
Cronbach’s alpha’s for “risk taking”, which is our main experimental variable was good (? = .82). The
same counts for tolerance for “ambiguity” (? = .75). The Cronbach’s alpha’s for self-efficacy and locus of
control however were somewhat disappointing (? = .52 and ? = .54 respectively).
In the next section both team and individual performance will be evaluated by means of a quantitative
analysis and by the results on the evaluative questions afterwards. The results are reported over the
remaining 48 people (12 chains).

RESULTS
As might be expected, this study confirms the existences of the bullwhip effect during the different
rounds played. Figure 1 shows the average variance in demand for each of the positions in the supply
chain. For each position we distinguish between low, middle, and high risk taking. As is normal in this
experiment the average variance experienced upstream is higher for each upstream position. The variance
in demand for the retailers is very low as demand is basically constant and changes only once early in the
game.

Figure 1 – Bullwhip effect: average variance in demand against position in chain for three levels of risk
taking

As a first step in finding the influence of personality characteristics, we conducted an analysis of
correlations and in addition for different personality characteristics we investigated by means of
ANOVA’s (analyses of variance) differences between groups, based upon a certain difference in
personality (e.g. low vs. high risk-taking).

The effects of personality characteristics
As a first way for analysing the data we investigated the relationships between personality characteristics
and the main performance measures in the game: backorder costs and inventory costs. Here, we found
significant correlations between efficacy and inventory costs (r = .34; p < .05) and negative correlations
between ambiguity and backorder costs and (r = -.29; p < .05) and between risk taking and backorder
costs (r = -.38; p < .01). Each of these results indicates that individual characteristics influence the
performance of a supply chain. The correlations mentioned are for all different positions in the chain. We
also investigated if correlations exist for personality characteristics and performance measures for each of
R W D F
0
50
100
150
200
250
300
350
low
high
middle

406
the four roles. Here, we do in general find the same correlations, but significance is heavily hampered due
to the small sample sizes of 12. The correlations found, stimulate further investigation of the data to find
out if, and in what way personality characteristics influence decision and performance.

Risk taking behaviour
Each of the supply chains consists of low, middle or high risk taking persons. For each position, we
investigate by means of an ANOVA analysis if low, middle and high risk taking persons differ with
respect to decisions and resulting performance. As dependent variables we used “backorder costs” and
“inventory costs”. Only the significant results will be presented in tables.

Table 1 – ANOVA result with risk taking behaviour as independent variable and backorder costs as
dependent variable; and associated means
risk taking
F = 5.32; p < 0,01
backorder
costs (means)
low (n = 17) 2276,00
middle (n = 9) 1018,00
high (n = 21) 884,19
total (n = 47) 1413,23

Table 1 shows that low risk taking persons have more backorder costs than middle and high risk
taking persons. An explanation could be that low risk-takers were too cautious with their orders, which
resulted in a high amount of backorders.
Furthermore, we conducted another ANOVA by incorporating more detailed measures of performance:
the “orders placed”, “outstanding orders”, “inventory and backorder level”, and the “upstream shipments
in the pipeline”. The results of this analysis are however tentative, since the 40 periods per person were
treated as 40 independent measurements. Results show that low risk taking seems to be associated with
(irrespective the position) on average negative inventory (e.g. high backorder). Only the manufacturers
have a positive inventory but still lower then persons with a higher risk taking profile. The negative
inventory level is, as can be expected closely related to high “outstanding orders” across most positions.
For retailers with a low risk taking the average order is higher than for medium and high risk taking
persons. Overall, the suggestion supports the idea that personality characteristics matter.

Efficacy
People were divided into three groups (for each role separately) having low, middle, and high efficacy.
Efficacy can be described, as mentioned, as one’s believe in the ability to complete a given task. We
conducted the same analyses, using the same performance measures as in the previous subsection. The
first ANOVA using “backorder costs” and “inventory costs” as independent variables shows no
significant results (F = 0,72; p =0,49 and F = 0,92; p = 0,41 respectively). The means suggest that people
with low self efficacy have higher backorder costs, but lower inventory costs than people with high self-
efficacy.
The second ANOVA analysis shows significant differences between the three groups in a number of
the measures. More specifically, a low level of efficacy results for wholesalers and distributors in higher
average orders. For the retailers, however, a high level of efficacy results in a higher average of orders.
For the factories, the results are less clear. Again these results, although significant, should be taken with
caution.

Ambiguity
For ambiguity the same type of analyses were performed as in the above two subsections. Table 2 shows
that people who prefer unambiguous situations (low on ambiguity) have higher backorder costs than
people that like ambiguity. Perhaps that people who like ambiguity felt more comfortable in the game,
which can be considered as an ambiguous situation, and therefore had lower costs than the other two
groups.

407
Table 2 – ANOVA result with risk ambiguity as independent variable on backorder costs as dependent
variable; and associated means
ambiguity
F = 3,68; p = 0,05
backorder
costs (means)
low (n=18) 2102,56
middle (n=13) 1171,69
high (n=16) 834,00
Total (n=47) 1413,23

The second ANOVA analysis again finds significant differences on “orders placed”, “outstanding
orders”, “inventory and backorder level”, and the “upstream shipments in the pipeline”. Here for all four
the three groups (low, middle, high) for all positions do show a difference. For all positions in the supply
chain roles (except the factory) a low level of ambiguity results in on average higher orders. The ANOVA
also indicates that “outstanding orders” and “inventory and backorder level” are also different and higher
on average.

Locus of control
As a final personality characteristic we divided our sample into low, middle and high internal locus of
control. A low level can also be more or less associated with external locus of control. Only the results on
inventory costs showed borderline significance (F = 2,08; p = 0,14). It seems that people who have a low
internal locus of control tend to have more inventory costs (mean “low” = 2012,25) than people who have
more an internal locus of control orientation (mean “middle” = 1607,00; mean “high” = 1006,71)
The other ANOVA revealed that external locus of control can be associated with a higher level of the
average order size for each of the positions in the chain. Interestingly, the standard deviation for the
external locus of control orientation is much higher than in most of the above cases and higher than for
the more internal locus of control orientation individuals.
The above further indicates that personality characteristics do matter in supply chain decision
making and task performance. However, we need further analysis to find out if we can track these
differences in what is considered and taken into account in making the decisions in the Beer game.
Initially the members in a team were selected on the basis of their risk taking behaviour. The next section
will look at these different risk taking teams, aiming at finding explanations for the main decision: the
order to the upstream position.

FURTHER ANALYSIS OF RISK TAKING
The main focus of this paper is the influence of the personality characteristics: risk taking. We will first
further look at typical differences in performance and supply task performance across the three different
levels of risk taking (low, middle, high). Based upon this analysis we relate the main decision variable to
demand, pipeline information, backorder and inventory information to find if people with differences in
risk taking base their decisions upon different information or interpret information differently.

Exploratory findings
The main performance measures in the beer game are backorder and inventory costs. Figure 2 shows that
for all positions persons with a low level of risk taking have on average lower inventory costs and higher
backorder costs then persons with a middle or high level of risk taking. Whereas the results for the low
level are consistent the middle and high level do not totally match the pattern. We think that low risk
taking people probably do not immediately follow the changes in demand, as the risk takers seem to do.
The effect is that orders are lagging behind demand and backorder costs are higher. The effect is most
likely to be stronger for distributors and wholesalers as they experience also the effect of their upstream
low risk taking supplier, that have relatively low inventory and an order backlog.

408

Figure 2 – Costs for different level of risk taking and positions

Figure 3 below plots the position in the chain against the quotient of the variance of demand and
variance of orders. It might look surprising that this quotient is largest for the retailers (position 4) but one
should recall that the variance of the demand for the retailer is very low. The interesting point in Figure 2
is that it indicates how the level of risk taking influences behaviour. For retailers low risk taking results in
a larger bullwhip effect than for higher risk takers, while for the other positions the opposite effect can be
detected. The underlying reason might be that low risk takers are relatively slow in adapting their order
level to the increase in demand, and do react after some time period with relative large orders. The similar
behaviour in other positions (slow or limited reaction to changes in demand) seems better (from a point of
view of bullwhip reduction) in the other positions. Higher level risk takers react quicker, which is good in
the retail position, but amplifies variance at other positions in the chain

Figure 3 – Average quotient of variance of demand and variance of orders for positions and personality
0
500
1000
1500
2000
2500
3000
3500
4000
average
backorder
costs
R W D F
low middle high
average
backorder
costs
average
backorder
costs
average
backorder
costs
average
inventory
costs
average
inventory
costs
average
inventory
costs
average
inventory
costs
0
5
10
15
20
25
30
R W D F
low
middle
high

409
Regression analyses: what drives different personalities?
The main (in fact only) decision in the Beer game is the amount ordered in each period. The previous
section indicated differences for individuals with different level of risk taking and the above figures
indicate further differences. In this subsection we attempt to find causes for the amounts ordered. The
idea is to explain the amount ordered using as independent variable the information available for the
decision maker. Following earlier research (Sterman, 1989; Croson at al., 2004) we conduct regression
analysis at the individual level to determine if differences can be detected. Next, we compare for each of
the positions in the game and for each level of risk taking if results can be generalised.

Table 3 - Regression analyses on each retailer using orders as dependent variable and demand,
backorders, inventory, and shipment as predictor variables.
retailers
risk taking
stand. ?
demand
stand. ?
backorder
stand. ?
inventory
stand. ?
shipment
R adj.
R
2
F Sign.
1.low ,514 -,291 .78 .57 13.88 p < .001
2.low ,442 ,524 -,247 .91 .81 43.27 p < .001
3.low ,521 .67 .39 7.12 p < .001
4.low -.356 .70 .42 8.16 p < .001
5.middle ,219 ,51 -,430 .87 .74 29.07 p < .001
6.middle .40 .06 1.61 n.s.
7.middle -1,117 ,97 .90 .79 37.59 p < .001
8.high -,611 .63 .33 5.57 p < .01
9.high -,379 .63 .33 5.88 p < .01
10.high .69 .42 7.96 p < .001
11.high -,658 .67 .39 7.17 p < .001
12.high ,399 -,617 .83 .64 18.61 p < .001

Table 3 shows that retailers with a low risk orientation tend to take demand, backorders and
inventory level into account when making a decision about how much to order from their wholesaler.
High risk taking retailers however basically look at their inventory level in coming to a decision on what
to order, while ignoring demand and backorders. None of the players, except one, looks at the amount of
shipments.

Table 4 - Regression analyses on each wholesaler using orders as dependent variable and demand,
backorders, inventory, and shipment as predictor variables
wholesalers
risk taking
stand. ?
demand
stand. ?
backorder
stand. ?
inventory
stand. ?
shipment
R adj.
R
2
F Sign.
1.low ,437 ,535 .79 .58 14.28 p < .001
2.low ,95 .80 .60 15.83 p < .001
3.low ,395 ,334 -,433 .85 .69 22.55 p < .001
4.low -,54 -,975 ,597 .73 .48 9.85 p < .001
5.middle ,73 -,31 .89 .76 32.26 p < .001
6.middle ,435 ,35 -,37 .79 .57 14.02 p < .001
7.middle ,205 ,633 -,343 -,399 .89 .76 32.13 p < .001
8.high -.340 .55 .22 3.71 p < .05
9.high ,29 ,52 .84 .67 20.52 p < .01
10.high .40 .07 1.70 n.s.
11.high ,61 ,61 .85 .70 23.65 p < .001
12.high ,41 ,74 .92 .82 44.71 p < .001

Unfortunately Table 4 does not show the same pattern. Backorders become more important for both
groups and demand also becomes more important for the high risk persons. The fact that demand
becomes more important for wholesalers than for retailers is logical since the wholesalers have to deal
with more variation in demand. Again shipments do not play an important role for all of the players.

410
Table 5 - Regression analyses on each distributor using orders as dependent variable and demand,
backorders, inventory, and shipment as predictor variables
distributors
risk taking
stand. ?
demand
stand. ?
backorder
stand. ?
inventory
stand. ?
shipment
R adj. R
2
F Sign.
1.low ,611 ,295 -,436 .91 .81 42.65 p < .001
2.low ,823 .83 .65 19.22 p < .001
3.low ,651 ,415 .96 .92 111.71 p < .001
4.low -,473 ,591 .78 .57 13.65 p < .001
5.middle ,615 -,240 .90 .79 38.17 p < .001
6.middle ,747 .80 .60 15.39 p < .001
7.middle ,963 -,505 .77 .55 12.82 p < .001
8.high ,225 ,333 -,466 ,358 .87 .73 26.54 p < .001
9.high ,773 .88 .75 29.63 p < .001
10.high ,383 -,530 .65 .35 6.26 p < .01
11.high ,388 -,305 .75 .51 11.05 p < .001
12.high ,381 ,774 -,357 .91 .80 40.19 p < .001

Compared to retailers and wholesalers, distributors face a lot more variation in demand. Table 5
shows that almost all distributors take demand into account. It looks like high risk takers emphasize their
inventory slightly more than low and middle risk takers. Backorders seem to be more taken into account
by low risk takers. For the distributors the shipments gain some importance.

Table 6 - Regression analyses on each factory using orders as dependent variable and demand,
backorders, inventory, and shipment as predictor variables
factories
risk
taking
stand. ?
demand
stand. ?
backorder
stand. ?
inventory
stand. ?
shipment
R adj. R
2
F Sign.
1.low ,68 -,19 .96 .92 110.77 p < .001
2.low ,42 ,34 ,42 .84 .66 20.28 p < .001
3.low ,648 .75 .52 11.37 p < .001
4.low ,76 .90 .79 37.17 p < .001
5.middle 1,12 -,37 .94 .87 65.99 p < .001
6.middle 1,12 .26 .93 .84 52.21 p < .001
7.middle ,840 .95 .89 78.60 p < .001
8.high ,58 -,26 .91 .82 44.46 p < .001
9.high .77 .55 13.01 p < .001
10.high ,73 -,36 .92 .83 49.56 p < .001
11.high ,65 -,30 .82 .64 18.40 p < .001
12.high ,92 ,55 -,45 .99 .97 320.42 p < .001

In Table 6 almost the same pattern can be identified. For low risk takers backorders seem to be more
important than for high risk takers, while high risk takers emphasize their inventory more than low and
middle risk takers. Also for factories shipments were hardly taken into account in the decision making
process.

DISCUSSION
This paper started with the idea that within the current experimental research on the bullwhip effect,
human behaviour has been misunderstood and undervalued. The previous section shows that differences
in behaviour for groups that differ with respect to a certain personality characteristic are tractable. The
ANOVA analyses show significant differences for all personality characteristics. These differences are
also tractable for risk taking behaviour which has been analysed in some more depth. A number of figures
indicate that low risk takers have higher backorder costs and lower inventory costs, while the high risk
takers show the opposite cost structure. We think that these differences are interesting, but an important

411
question arises: What causes these differences and can we understand how decision making in the beer
game context differs for different type of persons?
We seek to answer this type of questions by performing regression analysis to explain the order size
of different players focusing on the personality characteristic risk taking. In general low risk taking
persons seem to take into account back orders and demand more than high risk takers do. The last group
seems to base its ordering decision more on inventory levels. Both groups seem to not incorporate the
shipments into their decision making which indicates that type of information is hard to deal with, as
previous research already indicated. Compared to other players, retailers hardly take into account
demand, but that might be explained by the specific set up of our experiment (almost fixed demand). It
also seems that the type of decision behaviour also depends upon the position in the chain. Wholesalers
and distributors perform probably task that can be compared, but the role of retailer and manufactory
differs. While this is also clear from our experiments, it is hard to understand and label these differences
due to the small size of the sample. Moreover, the behaviour of persons also partly depends upon the
(quality of the) behaviour of the chain partners both in ordering and supplying. To better understand that
relationship in relation to personality, we will conduct analysis that controls for the effects of up- and
downstream behaviour.

CONCLUSION
This paper is a first attempt to better understand what has been labelled as ‘behavioural causes’ in the
supply chain management literature on bullwhip effects. The present paper is one of the first to
incorporate psychological constructs in the context of beer game experiments. One of our main findings is
that difference in personality characteristics do matter in performance of supply chain management tasks.
The paper attempts to explore what people with different personality characteristics decide and what the
differences are in their decision making processes. Some first ideas are formulated.
There are a number of limitations of this study. For an experimental study, the control over all
possible stimuli is not optimal. In order to simulate the dynamics of a real-life supply chain, we can
hardly control for the behaviour of players in the game. As a result, people face totally different situations
with regard to demand pattern and supply pattern. Future research can use computerised versions of the
game using a number of automated other players to control for those differences. We also aim for further
analysis of the available data, to develop measures for uncertainty experience in demand and supply.
Another limitation is the small sample. Future experiments can help in enlarging the sample size to
enable more analysis for separate groups. We think that the provisional findings (if confirmed by future
research) can help in improving supply chain management. One of the possibilities is to tailor training of
future managers to their personality characteristics. Another direction might be to present information in
such a way that planners use in the best possible way.

REFERENCES
Bandura, A. (1977), Self-efficacy: Towards a unifying theory of behavioural change, Psychological Review, 84, pp.
191-215.
Bandura, A. & E.A. Locke (2003), “Negative self-efficacy and goal effects revisited”. Journal of Applied Psychology,
Vol. 88, 1 , pp. 87- 99.
Barrick, M.R. & M.K. Mount (1991), “The Big Five personality dimensions and job performance: A meta-analysis”,
Personnel Psychology, 44, pp.1-26.
Croson, R. & K. Donohue (2002), “Experimental economics and supply-chain management - Experiments based on
the beer distribution game reveal managers' cognitive limitations”, Interfaces, Vol. 32, 5, pp. 74-82.
Croson, R., K. Donohue, E. Katok & J. Sterman (2004), “Order stability in supply chains: Coordination risk and the
role of coordination stock”, Massachusetts Institute of Technology (MIT), Working Paper Series, ESD-WP-2004-04.
Jacobs, F.R. (2000), “Playing the Beer Distribution Game over the internet”, Production and Operations Management,
Vol. 9, 1, pp. 31-39.
Lee, H. L., V. Padmanabhan & S. Whang (1997), “Information distortion in a supply chain: The bullwhip effect”,
Operations Research/Management Science, Vol. 37, 3, pp. 289-292.

412
Nauta, A., C.K.W. de Dreu, & T. van der Vaart (2002), Social value orientation, organizational goal concerns and
interdepartmental problem-solving behavior, Journal of Organizational Behaviour, Vol. 23, 2, pp. 199-214.
Nienhaus, J., A. Ziegenbein & C. Duijts (2002). “How human behaviour amplifies the bullwhip effect – a study based
on the beer distribution game online”, ETH, Zürich.
Rotter, J.B. (1990), “internal versus external control of reinforcement: A case history of a variable, American
Psychologist, Vol. 45, pp. 489-493.
Sterman, J.D. (1989), “Misperceptions of feedback in dynamic decision making”, Organizational Behavior and
Human Decision Processes, Vol. 43, 3, pp. 301-334.

doc_819372083.pdf
 

Attachments

Back
Top