Abstract:

Organizations pursue goals. Their performance in doing so is constrained by the available resources at each moment: strategy is about developing these resources into the future (“resource-based view” or RBV). However, human organizations are dynamic: not only does a manager try to influence the organization or resources in its environment – these resources also react to the decisions of other agents and other factors.
As recognized in “managerial and organizational cognition” (MOC), the decision makers’ understanding of a situation frames perception and decision policies; these frame decisions which lead to certain dynamics of resource accumulation and the according performance. Therefore the decision makers’ understanding or “mental models” are a relevant component for strategic management.

The field of “system dynamics” (SD) has contributed to “dynamic decision making”, the RBV and MOC by developing a structural theory of resource accumulation over time: resources (state variables) increase and drain via flow rates, which can be influenced by decision policies which are informed by the resource levels. Such are logically closed systems driven by interdependent sets of feedback loops.
Decision policies are designed by developing a simulation model validated against known behaviors and concepts, and then conducting scenario experiments. However, such model development requires previous training and substantial time and money to conduct. Therefore different attempts to provide learning by making learners interact with simulation models have been reported since the early ’90. Together with them the dispute about which kind of activity is required for sufficiently deep learning effects i son: modeling versus simulating.
One of the earliest simulation games in SD is the renowned “Beer Game”: players simulate the simplified operations of a beverage distribution chain from brewery to retailer over distributor and wholesaler. Even though the design is seemingly simple, the game is dynamically complex and players systematically generate the “bullwhip effect”.

However instructive this game may be, its strong simplifications lead to cognitive dissonances in many players and this hinders the transfer of the dynamic insights to the players’ own situations. Tis may be explained by the fact that the hypothetical conclusions reached at the end of the debriefing of the game are not developed into firm specifications which can be tested and validated in a simulation run.
This paper reports from an initiative to overcome this gap by using a model which reproduces the game’s structure and provides a set of typical decision policies to learners. The model is accompanied by introduction videos and readings. The learner is lead through the exploration of the typical decision policies under varying scenarios of demand and delivery delays. He has to develop his own hypotheses and select and combine the decision policies accordingly. This work is similar to the usual modeling activity, but does not require developing a full-blown simulation model. A simulated reference decision policy allows the learner to compare his performances.
By the end, the mental model of the learner has been informed by the confrontation with the decision policies. In order to validate this claim, learners go through a second “Beer Game” (on-line).

The model will be presented and an application with students from a bachelor in business informatics will be discussed.

Keywords:

Mental models, system dynamics, decision policies, beer game.