Introduction

Knowledge is considered by literature as one of the most important strategic resources for firms to be competitive. So, firms, especially operating in knowledge-based industry, need to create and manage knowledge. Accelerating economic, technological, social, and environmental changes enhance the dynamic complexity of the same systems, making difficult for managers to fully understand the behaviour of such systems, and so the knowledge management.

An effective decision making and learning in a world of growing dynamic complexity requires to expand the boundaries of our mental models and to develop tools to understand how the structure of complex systems creates their behaviour.

System Dynamics approach allows to model, describe, and understand the behaviour of complex systems, so improving the capacity of individuals and organizations to learn and manage the knowledge related to these systems.

Several authors used this approach to analyze some specific problems. Repening and Sterman (2001) use this technique to study the “improvement paradox”, in other words why, despite of the increasing number of tools and techniques available to improve performance, there is a little improvement in the ability of organizations to embed these innovations in their everyday activities. Black et. al. (2004) utilize a system dynamics model to explain why the implementation of new technologies often disrupts occupational roles in ways that delay the expected benefits.

In this paper we deal with the notion of System Dynamics, describe the software programs generally used to build System Dynamics models, and introduce an example of System Dynamics application, such as to evaluate to what extent knowledge can be considered as a main factor for firms agglomeration in industrial districts.

System dynamics

System dynamics is one approach to modeling the dynamics of complex systems such as population, ecological and economic systems, which usually interact strongly with each other. Systems Dynamics was founded in the early 1960s by Jay W. Forrester of the MIT Sloan School of Management with the establishment of the MIT System Dynamics Group.

The modelled systems are closed, governed by feedbacks (actions feed back on themselves, determining new situations that influence next decisions), and are history-dependent (the ending point of an action is highly dependent on the starting point, that, in its turn, depends on the previous action).

The tools used to model a system, utilizing system dynamics theory, are mainly two: causal loop diagrams (CLD) and stocks and flows (Sterman, 2000).

The former allows to describe a set of variables, linked by arrows, showing the causal influences between couples of them. To each causal link it is associated a polarity (positive or negative), expressing the influence of independent variable on the dependent one. The result is the creation of loops that may represent positive (reinforcing - R) or negative (balancing - B) feedbacks.

Casual loop diagrams are used to capture the hypothesis about the causes of dynamics and to show the feedbacks believed responsible for a specific problem.

However, causal diagrams do not allow to describe the stock and flow structure of a system. Stock and flows structures are utilized to describe variables that can be accumulated (stocks) and their rates (flows) of increasing and/or decreasing. Stocks represent the state of the system, and generate the information upon which decisions and actions are based. Stocks give systems inertia and create delays by accumulating the difference between the inflow to a process and its outflow. These accumulations can be tangible stocks, such as plant, cash, equipment, and intangible ones, such as employee skills, customer loyalty, knowledge, etc..

Then, a system dynamics model involves three different types of variables: stocks, which describe the state of the system, flows, which describe the rate of increasing/decreasing of the stocks, and auxiliary variables, which can be linked to stocks and flows and are used to better describe the system behaviour.

Software

There are four software programs that were designed to facilitate the building and use of System Dynamics models: Dynamo, iThink/Stella, PowerSim and Vensim.

• Dynamo: was the first system dynamics simulation language, and for a long time the language and the field were considered synonymous. Originally developed by Jack Pugh at MIT the language was made commercially available from Pugh-Roberts in the early 1960s. DYNAMO today runs on PC compatibles under Dos/Windows. It provides an equation based development environment for system dynamics models;
• iThink/Stella (www.iseesystems.com): originally introduced on the Macintosh in 1984, the Stella software provided a graphically oriented front end for the development of system dynamics models. The stock and flow diagrams, used in the system dynamics literature are directly supported with a series of tools supporting model development. Equation writing is done through dialog boxes accessible from the stock and flow diagrams. IThink is available for Macintosh and Windows computers.iThink guides business team through the creation of models that simulate business processes and scenarios;
• PowerSim (www.powersim.com): in the mid 1980s the Norwegian government sponsored research aimed at improving the quality of high school education using system dynamics models. This project resulted in the development of Mosaic, an object oriented system aimed primarily at the development of simulation based games for education. Powersim was later developed as a Windows based environment for the development of system dynamics models that also facilitates packaging as interactive games or learning environments. Powersim's modeling and simulation tools are used to map formal mental models into models that can be simulated and analyzed on computers;
• Vensim (www.vensim.com): originally developed in the mid 1980s for use in consulting projects Vensim was made commercially available in 1992. It is an integrated environment for the development and analysis of system dynamics models. Vensim runs on Windows and Macintosh computers. Vensim is used for developing, analysing, and packaging high quality dynamic feedback models.  Models are constructed graphically or in a text editor. Features include dynamic functions, subscripting (arrays), Monte Carlo sensitivity analysis, optimization, data handling, application interfaces, and much more.

An example of application: modelling knowledge transfer in industrial districts

The central proposition of the resource-based view theory is that firms select actions that best build on and maintain their unique set of resources in order to be competitive. In a world characterized by increasing uncertainty resulting from fast-changing technologies and global competition, it seems particularly crucial to identify new sources of competitive advantages for companies and regions. Innovation studies and cognitive science recognize knowledge as the main strategic resource to be developed or acquired in order to create new products and processes. So, firms, especially operating in the knowledge-based industry, need to create new knowledge and to learn for establishing competitive advantage.

In fact, as stressed in the recent strategic management literature, the competitive strategy cannot be based only on two sources of competitive advantage (cost and differentiation) but should be devoted to create and manage new knowledge, so fostering innovation. This new knowledge results as the collection of pieces of information and knowledge that are owned by a variety of parties, and then requires, to be developed, the combination of the external learning processes (e.g. learning by imitation and learning by interaction) with the internal ones (R&D activities, learning by doing and by using). In fact, the innovation process of firms can be conceived as an open system where heterogeneous inputs (internal and external knowledge) are transformed into outputs (results of innovation).

The paradox that characterizes the recent competitive scenario is that, although the market and the economy are more and more globalized, the localization of firms is still a relevant factor to achieve and maintain competitive advantage in the long-term. In fact, co-localized firms, such as industrial districts, benefit from both economic advantages (such as the reduction of transaction costs and the pooling of common factors of production), and the easy and fast knowledge transfer.

An industrial district is a “a sizable and spatially delimited area of trade-oriented economic activity which has a distinctive economic specialization, be it resource related, manufacturing, or services” (Markusen, 1996, pp. 296). Industrial districts provide an interesting example of local systems of production, that exploit benefits arising from agglomeration externalities, both economic and cognitive.

In an economic context greatly characterized by the increasing role that knowledge plays as a strategic resource for firms and regions competitiveness, system dynamics theory can be used to evaluate to what extent knowledge can be considered as a main factor for firms agglomeration in industrial districts.

In fact, developing a system dynamics model it is possible to analyse how the number of firms in an industrial district, then the agglomeration process, and its competitiveness it is affected by the knowledge created and shared inside the district, that can be considered as indicators of district/cluster attractiveness.

A possible representation

Influence among knowledge stock inside the cluster, knowledge stock outside the cluster,
and agglomeration process.

In the above figure, a possible representation of a System Dynamics model, which describes how knowledge affects the  firms agglomeration processes in an industrial district, is shown.

In this model we have introduced four stock variables:

• number of firms inside the cluster: it represents the agglomeration degree of firms. It is positively influenced by the firms entering the cluster and negatively by the outgoing ones;
• number of firms outside the cluster: the definition of this variable has been made necessary by the hypothesis of closed system, meaning that the total number of considered firms is defined. It is complementary to the number of firms within the cluster and characterized by the same flows of firms, with opposite directions;
• knowledge stock inside the cluster: it measures the different kinds of knowledge held by firms within the cluster. It is the total amount of firms knowledge stocks, considering only one time the knowledge owned by more than one firm, and it is positively affected by the internal learning rate;
• knowledge stock outside the cluster: it depends on the different kinds of knowledge held by firms outside the cluster, and it is positively affected by the external learning rate,

four flow variables:

• ingoing firms: it is the flow of firms entering the cluster, and it is positively affected by the cluster attractiveness;
• outgoing firms: it is the flow of firms going out the cluster, and it is positively affected by the cluster repulsiveness;
• internal learning rate: it is the flow feeding the knowledge stock inside the cluster. It represents the new knowledge learned by the cluster per unit of time, and it is positively affected by the number of contacts inside the cluster;
• external learning rate: it is the flow feeding the knowledge stock outside the cluster. It represents the new knowledge learned by firms outside the cluster per unit of time, and it is positively affected by the number of contacts outside the cluster,

and four auxiliary variables:

• cluster attractiveness: it represents the ability of the cluster to create new knowledge, and so to attract firms; it is positively affected by the knowledge stock inside the cluster, and negatively by the knowledge stock outside the cluster;
• cluster repulsiveness: it represents the ability of the environment external to the cluster to create new knowledge, and so to induce firms to leave the clusters; it is positively affected by the knowledge stock outside the cluster, and negatively by the knowledge stock inside the cluster;
• number of contacts inside the cluster: it represents the contacts that take place among firms inside the cluster, and it is positively affected by the number of firms inside the cluster;
• number of contacts outside the cluster: it represents the contacts that take place among firms outside the cluster, and it is positively affected by the number of firms outside the cluster.

An increase of the knowledge stock inside the cluster determines a growth of the attractiveness, with consequent increases of the incoming firms, of the number of contacts inside the cluster, and of the internal learning rate. The result is the creation of a reinforcing loop. On the other hand, an increase of the knowledge stock outside the cluster determines a reduction of attractiveness, with consequent decreases of  the number of incoming firms, of contacts inside the cluster and of the internal learning rate. The result is a balancing loop.

Similar dynamics characterize the links between the cluster repulsiveness and the knowledge stocks inside and outside the cluster.

References

Black, L.J., Carlile, R.P., Repenning, N.P. (2004) A dynamic theory of expertise and occupational boundaries in new technology implementation: building on Barley’s study of ICT scanning, Administrative Science Quarterly, Vol. 49, No. 4, pp. 572-607.

Markusen, A. (1996) Sticky places in slippery space: a typology of industrial districts, Economic Geography, Vol. 72, No. 3, pp. 293-314.

Repenning, N.P., Sterman, J.D. (2001) Nobody ever gets credit for fixing problems that never happened: creating and sustaining process improvement, California Management Review, 43(4), pp. 64-88.

Sterman, J.D. (2000) Business Dynamics – Systems thinking and modelling for a complex world (Irwin McGraw-Hill).

Spread the word!