SOME MODIFICATIONS INTRODUCED TO IMPROVE THE BEER GAME
José A.D. Machuca, Miguel A.D. Machuca, Angel Maresca
GILD.E.A.O.Dpto. de Economia Financiera y Direccién de Operaciones, Universidad de Sevilla
Avda. Ramon y Cajal, 1, 41018 Sevilla, Spain
ABSTRACT
The Beer Game is still, today, one of the tools with the greatest impact in.demonstrating that the
behaviour of a system is generated by its structure. However, we believe that in its original form
too much time is needed to play it and carry out a proper debriefing. In addition, it is not always
easy to guarantee the hypothesis of isolation for the different positions within the game.
Finally, we feel that participants often have difficulty in picking up quickly and clearly the
Process characterizing the game. To deal with these and other problems we have introduced
certain modifications which, in our view, totally or partially resolve these difficulties.
. Introduction
As is well known, the beer game offers, with great clarity and impact, a means to demonstrate
that the structure of a system generates its behaviour and to make management realize the
need to use Systems Thinking when dealing with complex organizations. On account of this,
we have been using it as a "shock weapon" to introduce courses in Systems Thinking and/or
System Dynamics. In order to improve the working of it in this context, we have gradually
incorporated a series of modifications (Machuca, J.A.D. et al., 1991), with a view to:
* safeguarding the observation of the rules of the game
* speeding up its progress
* reducing possible calculation errors
* reducing running time
* facilitating the debriefing
2. Some "physical" changes in the Beer Game.
In order to reproduce the classic isolation that characterizes the analytical approach, one of the
rules of the beer game is the prohibition of communication between the different players.
Therefore, no information should be available on the state of the inventory of adjoining
positions. Nor should the value of the orders placed by the player immediately to the left be
known until after the two delay periods which are a feature of the transmission of information
within the system.
However, the above-mentioned rule is not easily observed. The cases of beer moving along
the distribution chain are represented one-to-one by coins or chips which are placed on the flat
surfaces symbolizing warehouses and shipping delays; these can be seen easily, given the
physical proximity between players. In addition, the 1/1 correspondence between cases and
coins presents a further drawback. When large quantities aré being dealt with, which is not
infrequent, a real physical problem of snowing under" arises and, in addition, when real stocks
must be noted the count is slow.
In order to prevent these problems we decided on a number of changes. Firstly, we introduced
chips of various colours and sizes representing different quantities (1/5/10) so as to reduce the
number of chips in the chain. In addition, in order to facilitate the exchange of chips, we placed
308 SYSTEM DYNAMICS '93,
plastic tumblers in front of each player, containing sufficient quantities of chips of different
values. Secondly, at the pints representing shipping delays we placed opaque plastic tumblers
to represent haulage trucks; this eases the flow of movement of material and makes it
impossible to see the goods being transported. We also placed plastic tumblers on the
warehouses (one for each type of chip); in this way the beer in store remained concealed from
curious eyes.
As regards orders placed and incoming orders, these are noted down on slips of paper through
which the amounts written down could quite. often be seen. To prevent this, on the blocks
representing these orders we placed plastic tumblers into which the folded slips are put.
These modifications safeguard the isolation of each position and:at the same time speed up
the flow of material and information throughout the production-distribution chain.
As regards the gameboard itself, we carried out some changes here as well (see figure 1). We
decided to standardize the different positions; thus, we replaced "production requests” by
“orders placed” to bring it in line with the other positions. To complete this standardization we
introduced a block of incoming orders, to complement the previous block. These orders are
filled, whenever possible; by the assistants of the game's facilitator; inthis way the player in the
"brewery store” position need not know that there is no limit to the incoming supply, or concern
himself with filling the orders which he himself places.
With the above changes we were able to avoid the frequent doubts and breaks arising in the
early phases of the game, particularly from the players occupying the factory position.
Finally, we broke the board down into 12 rectangular parts (four of them 28 cm. x 20 cm. and
eight 28 cm. x 40 cm.) which can be fitted together. The arrows indicating the direction of flows
in the system can have two different lengths (the greater of them involving three: additional
rectangular blocks), thus enabling the distance between players to be increased or reduced as
required and also making the game adaptable to different table sizes.
3. Modifications in the prompts indicating the different steps. of
the game to the players
In the course of numerous repetitions of the game we observed that, particularly in the initial
phases (the first 8 to 12 weeks), some misunderstandings and doubts arose as a result of some
players having difficulty in grasping the mechanics of the game. This leads to the game being
interrupted and holds back its progress.
Working from the difficulties observed and discussions with the players, we decided to
introduce some slight modifications in the prompts directing the various stages of the game
and also clarify some of them or insist on certain aspects which were at times being ignored by
the players, leading to interruptions and/or errors. All this is projected on a screen throughout
the game (see figure 2).
4. Some modificatons in the Record Sheets
The original Record Sheets only include the concepts of “current stock", "backlog’ and “orders
placed". However, we found that players constantly need and generate complementary
information records such as "incoming orders" or “orders information to be filled". Accordingly,
and following the order laid down by the steps of the game (see section 3), we altered the
record sheet, including useful new concepts such as formulas for the calculation of orders to
be filled and backlog (see figure 3), which at times are forgotten by the players (1).
SYSTEM DYNAMICS '93. 309
is
WAVE Ky 2 ERED LY
DONqDY
ueospuly
DyudId
SUED 490g 84 $0 pieog eu *, aunBi4
SYSTEM DYNAMICS '93
310
STEPS OF THE BEER GAME STEPS OF THE BEER GAME
1. Receive inventory and avance 1. Move the contents of the beer trucks
shipping delays and receive inventory
Note down inventory received
2. Fill orders 2. Fill orders
Note down incoming orders
[os | a Backlog of orders
= F-|trom previous pric
3. Record inventory or backlog [eecmnees |= MIN(["r"] [Secs |)
lventory
Note down current inventory or backlog
4, Advance the order slip [=] = bs
3. Move orders placed to incoming orders
5. Place orders 4, Place orders and note then down
2a. Original text 2b. Modified text
figure 2. Steps in the Beer Game
We also created two sheets to draw other useful complementary graphics. On the first are
drawn the incoming inventory and incoming for each position; on the second, once the game is
over, the participants -except the retailer, who is the only one who knows it- draw the demand
from clients as they imagine it to be. This graphic was already envisaged in the original version
of the game for comparison, in the debriefing, with the real demand. Finally, we added a series
of “comment sheets” on which the participants are to note down impressions, changes in
decision-making , causes, etc., as the game progresses, indicating the period in which they
occur. This is very useful for the debriefing and the later in-depth study of the players’
behaviour in the course of the game.
5. Implementation of the game on a spreadsheet.
The mechanics of the game require a series of calculations to be carried out to keep up to date
the values of the different variables that come into play (inventory, backlog of orders, orders to
fill, etc.). The speed which is part of the dynamics of the game, together with a faulty perception
on the part of some players of some of the concepts used (e.g. backlog of orders, difference
between orders to be filled and orders filled, etc.) give rise to calculation errors and delays in
the progress of the game, due to breaks requested to clarify doubts or to correct errors that
‘have been discovered. In addition, once the game is over, the calculation of costs and the
creation of graphics for the results is also tedious and time-consuming and breaks the rhythm
generated by the game itself.
SYSTEM DYNAMICS '93 311
Record Sheet nt 1 Give
Position in the Company:
eek REEVED: RVENTORY RCOMNG ORDERS ["oapeasTomULiy | CURRENT AVERT BACKLOG) "ORDERS PLACED
-(1) Orders to fill (i) = Incoming orders (i) + Backlog (1-1)
(2) Backlog (i)= Orders to fill (i) - Orders really filled (i)
Figure 3 Record Sheet
In view of the above, using the Excel spreadsheet we created a representative model of the
game, which automatically works out for each team all the necessary calculations for the
different positions. Each player simply has to feed in the orders placed during each period.
(Figure 4) *
Period RECEIVED. INORDERS © ORDERS. BACKLOG CURRENT. ~». ORDERS
INVENTORY TOFILL INVENTORY PLACED
1 4 4 4 0 12 4
Qt 4 4 4 0 12 4
3 4 4 4 0 12 4
Figure 4: Spreadsheet. Calculations for retailer.
In another module of the spreadsheet the various costs generated in each position for every
period are calculated, as well as the total cost for each team.
The evolution-of the effective stock of the different positions 'in each team (real stock on the
positive side of the ordinates and backlog on the negative) is also obtained in graphic form,
both separately and superimposed in one graphic. The latter makes it possible to clearly
observe the phenomena of oscilation, phase lag and amplification characteristic of the game.
The same procedure is followed as regards the “orders placed” variable. The final series of
graphics covers partial and total costs. (Fig. 5)
To make for easier comparison of the evolution of the above-mentioned variables for the
different positions, it is sometimes advisable to smooth off the variables in order to reflect the
312 SYSTEM DYNAMICS '93.
trends more clearly. An adjustment was therefore incorporated, yielding reasonably satisfactory
results. (Figure 6)
STOCK EFECTIVO
600
500
400
300
200
100
o
100
200
200
400
ST EF MINOR. ST EF CON ST. EFDELG ST EF. FABRICA
PED EMITIDOS
160
140
120
100
80
cy
40
20
°
1 9 49
FABRICA
costs
25000
20000
i Faprica
15000 Ho | eo
Dloonceso
10000 t_
Hi inor
5000
° Ps
ree ee reper PA RNS ABB BBS
Figure 5. Graphics for the Beer Game (effective stock, orders placed, costs)
SYSTEM DYNAMICS '93, 313
PED EMITIDOS
seeneuncesorsoneees MINORISTA,
_ererrornemorerer CONCESION..
"DELEG.
—"FABRICA
1 9 Ww 25 33 4 49
Figure 6a. Original Graphic
PED EMITIDOS ALISADOS
18,00
16,00
14,00
-ceeeeecsooooesener ALTS MINOR.
12,00
10,00 erro: ALIS CONCES
8,00
6,00
ALIS DELG
nn ALIS FABRICA
4,00
2,00
0,00
1g ” 2s Ex} 41 49
Figure 6b. Smoothed Graphic
Figure 6.Smoothing off a series of values in the Beer Game (orders placed)
The software mentioned above can be used during the whole game (depending on the
proportion of computers to positions), thus making full use of its potential, since it is then
unnecessary to fill in the record sheets or make calculations, therefore preventing the resulting
errors. However, we generally use it when the game is over, to calculate costs and create
graphics, which-are projected onto a screen so that a number of interesting comparisons can
be made, such as:
* costs of the different teams
* evolution of teams' effective stock/ client demand
* evolution of effective stock/ evolution of orders placed by each team
* evolution of team variables/ evolution of "no strategy” variables (see section 6).
314 SYSTEM DYNAMICS '93,
6. incorporation of "no strategy" strategy into the debriefing
When the game is over and the various costs and graphics are obtained for each team, the
facilitator conducts the debriefing so that the participants can see that, contrary to what they
might think:
a) there is no one "culprit" in each team.
b) neither are external events the reason for the behaviour obtained.
Having reached this point, as P. Senge (1990, chapter 3) points out, the players have only one
recourse to fall back on: to blame the system and say it is probably unmanageable. To
demonstrate that this is not true and that a better job can be done, the author puts forward the
example of "no strategy" strategy, that is, each player places orders exactly equalling those
coming in, without concerning himself with anything else. In following these guidelines a
certain stability is reached around week 11, and although it is, of course, not a good strategy, it
serves to show that’ the huge oscilations, characteristic of the game (Figure 7) can be
eliminated, better costs results being obtained in 75% of cases than were obtained by the
different teams (see Figure 8).
EDIDOS EMITIDOS
8
7
6
s
4
3
2
1
o
1 9 7 25 3 a 0
MINORISTA CONCESION. DELEG. FABRICA
STOCK EFECTIVO
2 26 31 6 4 46
ast
‘ST EF MINOR. ‘ST EF CON ST. ERDELG ‘ST EF. FABRICA
Figure 7. Results of the "no strategy" strategy (orders placed, effective stock).
When faced with these data, participants see the last possible excuse for generally deplorable
results disappear into thin air, for they now see that:
SYSTEM DYNAMICS '93 315
c) the system is not unmanageable,
which brings down the last barrier between them and the final conclusion: It is the structure
of the system that generates Its. behaviour.
In view of the above, we believe that it was advisable to somehow introduce the "no strategy”
strategy into the game. Depending on the case, we do this in two different ways:
MORETTI TEAM COSTS
25000
20000
Tl FABRICA
15000 Baca
Ocaceso
10000 a
Hi minor
5000
0
“NO STRATEGY” STRATEGY COSTS
x = ©
y a 2 2
©
Miminon Dloonceso Bi paEG Il FABRICA
Figure 8. Costs of the “no strategy” strategy compared with typical game costs
* during the game (when there is a big enough number of participants), making a team
Play this strategy.
316 SYSTEM DYNAMICS '93
* after the game, once this point has been reached ‘in the debriefing, through the
spreadsheet-based program mentioned in the previous section.
We reckon that the results which derive from it make it worthwhile to incorporate this strategy
into the game and use it in the debriefing.
7. Documentation produced for the Beer Game.
In order to make the task of potential Beer Game facilitators easier, and following the lines and
objectives mentioned in another paper presented by us at this Conference (Machuca, J.A.D. et.
al, 1993), we created comprehensive documentation (2) (Machuca, J.A.D. et al., 1991)
consisting of: -
a) Player's Manual
b) Teacher's Manual
¢) Set of 31 transparencies complementing the Teacher's Manual.
In addition to the above, we have produced the following software:
d) Presentation for Macintosh to replace the transparencies (1992).
e) Spreadsheet for calculation of the results of the beer game and creation of the
relevant graphics (see section 4) (1991).
Notes
(1) These formulas are also permanently projected into a screen from the transparency showing
the steps of the game (see section 3).
(2) The Player's Manual and the Teacher's Manual are based on our own experience (see the
Present paper) and the following works:
* Senge, P.: "Outline for postgame discussion". M.I.T. D-3236
* Senge, P.: “The fifth discipline". Ed. Doubleday Currency, 1990.
* Sterman, J.:"Instructions for Running the Beer distribution Game", M.1.T. D-3679,
1984.
* Sterman, J.:"Modelling managerial behavior. Misperceptions of feedback in a Dynamic
Decision making experiment", Management Science, Vol. 35, n° 3, 1989
References
* Machuca José.A.D.: "Documentaci6n para el Juego de la Cerveza", DEFDO, 1991
* Machuca José A.D, Machuca Miguel A.D., Ruiz José C., Ruiz Antonio:"Systems Thinking
Learning for Management Education. What are we doing in Sevilla?", in Machuca J.A.D. and
Zepeda E. (Eds), System Dynamics'93: The role. of Strategic Modelling in International
Competitiviness, The System Dynamics Society, 1993
* Senge P.: “The fifth discilpine", Doubleday Currency, 1990
SYSTEM DYNAMICS '93 317,
