The Reality of the Beer Game

[elsd@sepa.tudelft.nl]

Another beer game story:

Two weeks ago I played the beer game with employees of a production company
in The Netherlands. People from different departments participated in the
game. The logistics manager had asked to play the game, as the company had
just experienced a situation where the supplier had not been able to deliver
parts as quickly as the clients had expected, and consequently the
production companys clients had ordered more, without the production
company realising this. This meant that they now have a warehouse full of
stock. A situation that may last until October. The logistics manager wanted
the game to be played in order for other departments of the company to
understand what was going on, and to understand why he e.g. wants to
decrease time delays and would like more client information/communication.
The companys employees thought that the beer game situation was very
recognisable, and the production company is also going to play the beer game
with one of their largest clients this afternoon.


Regards,
Els van Daalen
Delft University of Technology
The Netherlands
elsd@sepa.tudelft.nl

[Scott Rockart]

Professor Coyles post suggested that many people may be
interested in seeing documentation for Beer Game behavior in the real
world. Professor Jan Hammond at Harvard provides perhaps the most thorough
and compelling documentation of a single industry example in her case study
of Barrilla SpA (HBS case #9-694-046, parts A through D each with slightly
different numbering). She presents data on the behavior of the supply
chain at for Italys largest pasta company (oscillation and amplification
up a four-stage supply chain) and the steps taken to reduce the
oscillation. John Stermans new book Business Dynamics also has a series
of great examples on pages 664-667 from a variety of industries.
Professor Coyles post also reminds us that the Beer Game is a
simplification of the physical and decision making elements of the real
system when he says "A real brewery does not oscillate explosively and
would go
bankrupt if it did." In most supply chains there are physical capacity
constraints (e.g., the number of vats in the factory, the number of trucks,
or the refrigerator space) that act as non-linear bounds on oscillation.
The constraints, as Barillas example shows, still leave room for costly
oscillation especially since managers often set capacity to accommodate the
perceived demand cycle.
It is also true that in real systems managers may use more
sophisticated ordering policies than proportional control. As Forrester
points out, and as can be seen in the game, many of these policies can be
self-validating and serve to sustain oscillation (Appendix L in Industrial
Dynamics) as firms build up inventory for expected peaks then market
aggressively to get rid of that excess. The rules can also amplify
oscillation further as teams project the current trend outward (watch a
team that tries to forecast demand). While I am not a supply chain expert,
I believe that even managers in organizations armed with excellent
information systems to track point-of-purchase data have to deal with
issues around trusting the information over the demands of less well
informed customers.
Most importantly, oscillation is not an outcome of proportional
control properly applied. Oscillation only occurs in the game because
people forget how much they ordered in the past. If teams incorporated
their past orders into their proportional control rule the system would
overshoot but then fall to a stable level when hit with the step in orders.
There would be no unexpected arrival of excess inventory to unload and thus
no oscillation. Players that return as "experts" often prove Professor
Coyles second to last point "the lessons it teaches are not evident" and I
am eager to read Ed Andersons and D.J. Morrices contribution to the
literature.

From: Scott Rockart <srockart@MIT.EDU>
==========================================================================
Scott Rockart
Sloan School of Management
Massachusetts Institute of Technology
50 Memorial Drive, E52-511
Cambridge, MA 02142 USA
Tel (617) 253-4023 Fax (617) 253-2660
==========================================================================

["Keith Linard"]

As a teaching tool, the beer game is magnificent. After playing the game I
build with the students an uncontrolled model ... and of course we get
increasingly dramatic oscillations. Then come the lessons:

1. To what extent is the behaviour due to use of Euler rather than
Runge-Cutta X? (Judiciously chose time steps & delay ensure that this is a
significant factor).

2. Do stocks go negative? Is this because outflow exceeds stock ... what
decision rule does this imply? In what circumstances might this be valid /
invalid? Or is it because inflow goes negative ... what decision rule does
this imply ... etc

3. If stock may not go negative, what decision rule achieves this? Limit
function ? Some complex "If statement" (at this point introduce Boolean
statements as alternatives to "If").

4. What about our ordering rule & what is the appropriate performance
indicator to inform the ordering process?

5. In real life, what other controls stop a runaway growth in oscillations
(if nothing else, finance limits on clerks etc).

As for real life examples of "beer game" type behaviour ... office
accommodation & housing markets in Canberra have seen 4 booms-and-busts
since 1976. More spectactular has been the pattern of Army recruiting in
Australia over the past decade (out of a total full time Service of about
26000): 300, 1000, 3000, 4000, 1000, 400 ... which oscillations will go
through every Army training centre for the next 10 years or so.

Keith Linard
University of New South Wales
Centre for Business Dynamics & Knowledge Management
From: "Keith Linard" <k-linard@adfa.edu.au>

[Nelson Repenning]

I dont think anybody claims that the beer game captures the dynamics
normally associated with brewery supply chains. I believe it has it origin
in Professor Forresters initial studies of General Electrics
difficulties with instability in its refrigerator manufacturing plants. In
fact the game was originally called the "refrigerator game" and Im not
sure who had the stroke of spin genius to rename it the beer game.

at any rate, at least at MIT, we do not use the beer game to teach people
about supply chain dynamics or proportional control. Instead, we use it as
a very compelling and hands-on way to teach the foundational concepts of
system dynamics, mainly that structure creates behavior, that all decisions
are made based on mental models (which are often flawed), and that, while
the system provides powerful data suggesting a particular person or set of
people is to blame for the poor performance, this is typically a very low
leverage approach to solving problems in complex systems.

Also, there are some supply chains to that do exhibit this kind of
oscillatory behavior. Jan Hammond at Harvard Business School has written a
nice case on the Barillia pasta company that experienced significant
volatility in its manufacturing orders, despite the fact that the consumer
demand for pasta (in Italy) was very steady.

so despite some limitations, I find that the beer game remains one of the
most important tools in my SD teachers tool kit.


Also for those interested, I believe Roberta Spencer at the System Dynamics
Society sells are the relevant materials for delivering the game.


nelson



--------------------------------------------------------------
Nelson P. Repenning
Robert N. Noyce Career Development Assistant Professor
Operations Management/System Dynamics Group
Alfred P. Sloan School of Management, MIT
30 Wadsworth St, E53-339
Cambridge, MA 02139
phone: 617258-6889 fax:617258-7579 e-mail:nelsonr@mit.edu
http://web.mit.edu/nelsonr/www

["Jim Hines"]

Just a couple of anecdotes on the reality of beer games.

A group of executives at a large chain of supermarkets once played the beer
game. At the point in the debrief when folks realized that placing multiple
orders for a given inventory shortage was the root of the problem, a woman
stood up and said that that the company was currently making the same
mistake in hundreds of locations. She said she had just returned from
traveling the country, trying to get store managers not to over-order. She
reported she had met with mixed success at best in explaining the problem to
store managers. She also said that its very difficult **not** to
over-order, even if you understand the problem. When you are out of
product, everyone expects you to DO something. About the only thing a store
manager can do is place another order.

I once interviewed a commercial real estate developer. Real estate has
beer-game type oscillations, as Keith Linnard just pointed out. Normally,
it wouldnt be surprising to see these oscillations at the industry level,
where one firm may not know that other firms are also responding to
inadequate supply (indicated by high prices). But, in building office
towers, the information is right there: Permits are public. And, even if
developers are illiterate, they can ride around the city and count the
number of holes being dug. I asked the developer with my usual tact, "Are
you guys stupid or what?". The developer paused only for a moment and said,
"Were stupid". Apparently, if you are a developer, you develop when you
can, because developing is what you **do**. And, you can get financing when
prices are high -- even though building itself takes years and other
buildings may also be going up.

Regards,
Jim Hines
From: "Jim Hines" <jhines@MIT.EDU>
MIT

[Bill Harris]

Great anecdotes, Jim. Here are three more notes on the "reality of
Beer Games":

1: With a partner, I once led a group of manufacturing people (who
should know better, right?) in the Beer Game. We used pennies as
markers, and we thought we were more than generous in our
estimates of the number needed. Part way through the game, we
began to see more pennies in piles than we had ever seen, and we
had to make an emergency run to the local accounting department to
raid petty cash. (It was interesting to explain to accounting
_why_ we needed another bag of pennies.)

Peoples reactions were telling. Part way through (about when
they could look up the table and see armies of pennies massing for
attack, some in the game began to laugh and say, "Its all going
to fall apart now." They knew the pattern, but they were
powerless to stop it. That sounds a lot like the people in your
examples.

2: As Im sure others have noted, participants groused that the Beer
Game rules were fixed--you couldnt possibly win, nor would anyone
structure their supply chain that way. In a different session, we
added a second stage to the game. After playing it once by the
rules, watching everyone lose, and making those claims, we gave
them the right to rewrite the rules. Anything was fair, as long
as they wrote the rules down in advance.

While I forget what changes they made, they did rewrite the rules
to fit their insight (admittedly, they didnt have a lot of
time--I think we gave them an hour for redesign--and their
computing power was limited to calculators). We replayed the
game, and they did, well, _much worse_ than they had with the
original rules.

That helped us make our point: these sorts of systems arent
necessarily intuitive, and modeling and simulation (whether manual
or computerized) can be a great way to manage risk. I think we
closed by showing a computer model of the game and demonstrating
the speed and ease of changing the rules in that version to try
out new policies.

3: "Pipeline Inventory: The Missing Factor in Organizational Expense
Management" (National Productivity Review, Summer, 1999) makes the
claim that the expense management system in many companies is a
computerized version of the Beer Game. It describes an
organization that had significant variance fluctuations with a
period of perhaps 4 months; with revised feedback, those
fluctuations were cut by 90-95%. ("Applying System Dynamics to
Business: An Expense Management Example," available at
http://facilitatedsystems.com/pubs.html, describes a model of
that system.)

Regards,

Bill
--
Bill Harris 3217 102nd Place SE
Facilitated Systems Everett, WA 98208 USA
http://facilitatedsystems.com/ phone: +1 425 337-5541
From: Bill Harris <bill_harris@facilitatedsystems.com>

["Thompson, Jim B246"]

Twenty-five years ago, the printed circuit business was called "hi-tech",
and the company I worked for was the largest producer of printed circuits in
the world. The printed circuit business combined electrical, chemical and
mechanical engineering, so we had no shortage of people schooled in control
theory. We sold product to the electronics industry, everything from
consumer electronics to supercomputers.

>From the inception of the printed circuit business in the early 1950s,
investment analysts characterized it as high growth and highly cyclical.
There were 3 to 5 year periods of rapid sales growth followed by 3 to 5 year
periods of declining or negative growth rates.

Electronic component purchasing agents who had more than a few years of
experience would "sense a change" in the marketplace and, to avoid
shortages, would increase the size of requisitions from manufacturing. When
manufacturing schedules tightened, production engineers at our customers
would try to anticipate potential shortages or partial shipments of circuit
cards by placing orders for the same custom products with more than one
circuit manufacturer. This "double-ordering" was considered such a standard
practice that the printed circuit industry association had a regular
double-order report to its members.

Savvy purchasing agents at our customers became aware that their double
orders were being ignored, so they doubled their orders again. I could go
on, but you get the idea. When we ran our production, sales and general
managers through the beer game, they were always amazed to see how close to
reality it was.

If you would like concrete evidence of this story, read the sections on
inventory and working capital in the public reports of Cray, Burroughs,
Wang, Prime, Atari, Coleco, Osborne, Kaypro and other companies that
dominated the business headlines of that era.

Jim Thompson
CIGNA HealthCare
Phone: 860.726.8607
Fax: 860.726.7966
E-mail: jim.thompson@cigna.com