System Dynamics


Online Foresight Guide

System Dynamics is a tool to investigate and model complex dynamic problems in terms of stocks (the accumulation of things), flows (the motion of things) and feedback loops at any level of aggregation.

Overall description

System Dynamics is a tool to investigate and model complex dynamic problems in terms of stocks (the accumulation of things), flows (the motion of things) and feedback loops at any level of aggregation.

It was popularised with 'Limits to Growth', a summary of the results of a computer modelling exercise concerned with the future development of the world economy. Using relationships among the major variables, which included population, pollution, resources, capital and land, it purported to show that within the next 50 to 100 years the world system, as we now understand it, would collapse. The book became a best-seller, and formed a strong link in people's minds between systems ideas, computer modelling and predictions of future catastrophe.

The systematic approach gives the impression of objectivity but nevertheless, the variables are defined by people involved so that the method is as subjective as others.

When is this method appropriate?

System Dynamics is now one of the most commonly used forms of computer simulation for dealing with many faceted problems. The objective of this method is to find the conditions under which a system under study will evolve and in what direction. The tools used are models that represent symbolically the reality of the system. It aims at considering the interrelationships between the components of an organisation or environment rather than looking at each component in isolation.

It is usually developed using a specific computer language (though it is in principle possible to use spreadsheets and similar methods to perform the tasks involved.) A system dynamics application starts with the identification of a problem. The modellers should then draw in all major patterns of influence that together create the 'system' that produces the problem. A successful model is able to simulate these patterns and produce system behaviour. Different values for variables and different policy structures may then be introduced to simulate how the system would respond to different circumstances or initiatives. This method searches for the causes of system behaviour that lie within the system, with events 'outside' serving as triggers rather than causes.

Approach (Step-by-step-guide)

This method looks for dynamic patterns, and describes them in terms of structural relationships between their multiple positive and negative feedback loops and the levels and rates of the primary variables. The design of a system dynamics model begins with a time frame. The factors that contribute to the problem are listed and their structural relationships sketched with particular attention to characterising them as levels and rates that feed or drain them. Levels and rates need to alternate in the model; no level can control another without an intervening level. The next step is to quantify these factors and the assumptions behind them. Computer simulations can then be run to test the validity of the model. The model will begin from the initial quantified values for the variables and step through them at discrete time intervals. The basic computer model employs a set of first order, non-linear differential equations to reflect changes over time, with the chosen time interval small enough so that system behaviour appears continuous.

Resources needed (time, budget, labour force, skills)

This method requires considerable expertise in modelling and computation.

Outputs

The first level output can be qualitative synthesis which helps policy-makers ask better questions and may help anticipate patterns and sources of dysfunctions. However, the complex and dynamic patterns generated by information feedback and circular causality are difficult to utilise without computer support.

Pros and cons

Advantages:

System Dynamics models are used to understand and anticipate changes over time in puzzlingly complex systems. It can be used with what are thought to be 'data poor' problems. The information base for the conceptualisation and formulation of System Dynamics models is much broader than the numerical database employed in operations research and statistical modelling. This method can be useful to gain insight and understanding in a messy situation by sketching increasingly sophisticated causal loop diagrams.

Drawbacks:

A System Dynamics model is only capable of running one version of a situation at a time, although it may capture a great deal of variety in the changing values of its variables. Different stakeholders or groups with different cultural or political agenda might bring different assumptions and thus see a quite different picture. A system dynamics diagram can become very complex when actual situations with lots of variables are modelled.

Possible complementary methods

System dynamics can be used with most of other models to enhance understanding of system behaviour or to simulate the future.