Playing the beer distribution game over the Internet

Production and Operations Management, Spring 2000 by Jacobs, F Robert

PLAYING THE BEER DISTRIBUTION GAME OVER THE INTERNET*

This paper describes an Internet implementation of the Beer Distribution Game. Many teachers demonstrate the bullwhip effect that is often observed in supply chains by playing this game with their students. This implementation has the advantage of considerably reducing the time required to play the game.

(SUPPLY CHAIN MANAGEMENT, BEER GAME, INTERNET, GAME)

Introduction

Possibly one of the most widely used classroom exercises for demonstrating the dynamics of a supply chain is the Beer Distribution Game. The System Dynamics Group developed.the exercise at the Massachusetts Institute of Technology's Sloan School of Management (Sterman 1989). Normally the game is played manually on a game board with paper demand and order cards. Pennies are used to track the movement of cases of beer. This paper describes a version of the game that can be played over the Internet that has the advantages of quicker setup, quicker game play, and quicker analysis of game results.

The Beer Distribution Game simulates a phenomenon known as the "bullwhip" effect. The classic example of the bullwhip effect was observed at Procter & Gamble (P&G) with the sales of Pampers diaper (Lee, Padmanabhan, and Whang 1997). While the consumers, in this case babies, consumed diapers at a steady rate, the variability of demand grew as it progressed up the supply chain. For instance when P&G looked at demand for raw materials to their suppliers, such as 3M, they saw large swings. Many additional examples of the phenomena have been identified in the literature.

The manual version of the game is played on a board that represents the production and distribution of beer (see Figure 1). Teams of students represent different parts of the supply chain. Players take on the following roles to simulate the supply chain echelons for each brewery:

the retailer sells cases of beer to a consumer and orders cases of beer from the wholesaler,

the wholesaler sells cases of beer to the retailer and orders cases of beer from the distributor, and

the distributor sells cases of beer to the wholesaler and orders beer from the factory.

The factory brews the beer.

Pennies represent cases of beer and are moved between the positions on the board. The object of the game is to minimize two inventory related costs: holding cost ($0.50/case/ period) and backordering cost ($1.OO/case/period). Costs are assessed each period at each echelon as the game is played. During each period the players receive orders, evaluate their inventory position and decide orders and shipments for their echelon. Consumer demand for beer is simulated using a deck of cards according to a predetermined sequence and given to the retailer each period. A fixed shipping delay of two periods between each echelon simulates the time required to receive, process, ship and deliver orders. In the case of the factory, a lead time of two periods is required to produce a new beer order.

The game starts in equilibrium with 12 cases of beer in inventory at each echelon and 4 cases in each of the delay positions (see Figure 1). Normally, the simulation begins with four weeks of steady demand (4 cases per week) and all the players are directed to order and ship four cases each period, to maintain the initial equilibrium. Following the four-period startup, players are then instructed to order any quantity they wish. At this point, there is an increase in customer demand to eight cases per week. This change in demand induces disequilibrium into the system to which the students must react. A complete description of the game including the specific "rules of play" is given in Heineke and Meile (1995).

Note that the increase in demand is introduced at the retailer, who may respond with a change in the size of the order to the wholesaler. The retailer, in deciding what to order, may perceive the increase in demand in a number of ways. The wholesaler does not see the change in the order size until the next period. So the knowledge of this change in demand propagates through the system over the next four to five periods.

Sterman (1989) performed econometric tests to explain player behavior and found that an anchoring and adjustment heuristic for stock management was a good fit to the behavior. As noted by Sterman, players fall victim to several "misperceptions of feedback." Specifically, the players failed to account for control actions, which had been initiated but have not yet had their effect (i.e. they were looking at inventory on-hand rather than inventory position). In Sterman's studies, the majority of players attributed the dynamics they experienced to external events, when in fact these dynamics were internally generated by their own actions.

Professor Dan Steele of the University of South Carolina, has developed an interesting model of the process that the decision-maker uses in playing the Beer Game (see Figure 2). His model includes a forecast of the future demand. This forecast is used to calculate a stocking level goal that the player thinks is appropriate. An actual order is then placed in an attempt to bring the inventory up to this target level. When the upstream player sees this order, for example when the wholesaler sees the order from the retailer, the player reacts by ordering even more inventory. As we move up the supply chain toward the factory, the impact of the demand spike is further overstated, thus inducing the bullwhip effect.