I utilised the System dynamics modeling methodology to model the problem within the context of the environment that it resided in. The methodology enabled me to fully understand the system and its characteristics. It is important to note that System dynamics modeling is focused on modeling a problem versus modeling a system. There is a subtle difference in this statement.
Traditionally individuals would model a physical system and then use it to test various “what if”
scenarios. System dynamics on the other hand models a problem and considers both quantitative and qualitative or behavioural aspects. System dynamics recognizes that system behavior is not imposed from the outside but rather from within the boundaries of the problem and that the system behavior is a function of the interactions of the variables within the model (Richardson, 2011).
44 Another fundamental reason for applying System dynamics to this particular problem is that frequently one finds that individuals will follow policies, which they assume will lead to problem resolution. This is done in conjunction with the individual’s dependence on using their intuition or gut feel to determine solutions to complex behavior (Forrester, 1994). System dynamics enables an organisation to move away from trying to understand the impact of the individual on the system but rather to understand and test theories and policies on the system. It helps us explain the endogenous generation of macro behaviour from the myopic behavior of individuals (Sterman, 1989).
The sales and operations planning process that was the focus of this study is extremely complex, spans the entire supply chain including internal and external parties as well as having casual and feedback loops that spans across varying time frames. System dynamics uses stock-flow diagrams to show causal loops, which includes those causes that create the behavior of interest (Forrester, 1994). The applicability of utilizing the system dynamics methodology and thinking to this study is well supported by the literature.
The System dynamics methodology seeks to model social systems and problems across the short, medium and long term, using inputs from the mental database as the primary input (Forrester, 1986). System dynamics modeling hence seeks to convert qualitative data into quantitative data to build the model using the relevant software. In this instance, I used software called iThink. This software is specifically developed and used in System dynamics applications. It is therefore essential that the data is collected and converted ready for use in a robust manner to ensure that the model accurately depicts the problem and hence system.
A common pitfall of modeling using System dynamics is that practitioners sometimes over use causal diagrams beyond the limits of mental simulation. The use of computer hardware and software technology overcomes this particular obstacle (Richardson, 1999)
45 The methodology I followed is Sterman’s five step process. The five steps can be seen in the Figure below.
Figure 3.1: System Dynamics modeling process
(Sterman JD, 2000). Business Dynamics: Systems thinking and modeling for a complex world, Boston, Irwin McGraw Hill, Pg 87.
It is important to note that as can be seen in the centre of Figure 3.1, there is a network of lines joining each step to the other. This is indicative that the system dynamics methodology is an iterative one. The intention of this is purposeful, as during the modeling process the modeler is continuously improving his/her understanding of the situation and hence the model is improved.
Forrester (1986) states that the holistic understanding and building of the model is typically achieved by completing diagrams, understanding concepts, stock-flow diagrams and doing simulations which can be compared to the real world as a test. Hence as the modeler or team member one should not expect the five steps to follow in series but rather a number of iterations of the model and to move between steps until the problem is accurately modeled and adequately depicts reality. This must translate into the model being fit-for-purpose.
46 Problem articulation or boundary selection seeks to identify the issue or problem within a particular environment as well as the scope of factors involved. Boundary selection is a critical step as having to narrow boundaries or scope will result in certain insights being omitted whilst having too broad boundaries results in unnecessary noise and data collection. Both of which could result in a model that does not accurately depict the problem and hence cannot be used for the intended purpose.
Dynamic hypothesis is the step in which the modeler would list or sketch the interactions and feedback loops of the problem which would enable a comprehensive understanding of the problem and its drivers. This aids the modeler in understanding the problem and feedback loops. I would like to once again emphasis that this is an iterative process hence the understanding of the problem and complexities involved would not occur at the first attempt.
Formulation is the process in which the modeler transforms the hypothesis into detailed diagrams showing feedback loops and corresponding equations (Forrester, 1961). Stock-flow diagrams form a part of this step as well.
Testing simply put is the validation and verification process applied to the model built. Validation is the process followed to authenticate that the model was constructed in accordance to the prescribed methodology. Verification on the other hand is the process of comparing the models behaviour over time with evidence from the real world problem and environment. This is done with the intent of establishing the accuracy of the model and ensuring that the model depicts reality, is plausible and fit for its intended purpose. Depending on the results, the modeler may need to revisit steps 1, 2 or 3 or a combination thereof.
Policy formulation and evaluation is the final step and will only occur once the system dynamics practitioner is confident that the model accurately depicts the problem and is able to simulate the real world problem and behaviours. The system dynamics model is then used as an improvement and learning tool to evaluate policy improvement interventions.
Information collection during this process was acquired from three sources viz mental, written and numerical databases. However, the key source of information is from the mental database with the content of information decreasing as one goes from the mental, to written to numerical databases as illustrated in Figure 3.2 below. Qualitative data collected was transformed into a format relevant for use in the software specified.
47 Figure 3.2: Sources of information and relative quantity of information acquired per source.
Forrester, J.W. (1986). Lessons from system dynamics modeling. The 1986 International Conference of the System Dynamics Society. Sevilla, October, 1986.
It is clear from the above diagram that a high reliance is placed on the mental database for inputs in the model building process. Lune-Reyes & Andersen (2003) stated that qualitative analysis when done properly brings a high level of rigour and robustness to the model built.