The suitability of system dynamics for a specific problem is discussed as follows in Vennix (1996, p104) in a section entitled, "Initiation of a Project: The Suitability of System Dynamics." Vennix discusses four "guidelines on when to use system dynamics and when not."
1) "The problem needs to be dynamically complex."
2) "Typically system dynamics addresses long term problems, frequently those problems which have been tackled unsuccessfully in the past."
3) "Check whether it is possible to generate a reference mode of behavior which may represent a problem. If this cannot be accomplished one needs to be particularly careful."
4) "Check whether the problem description gives rise to thinking in terms of flow processes."
Summing up, Vennix writes, "System dynamics is primarily a diagnostic and impact assessment method: finding out what the problem is, what structural causes are responsible for it, and which policies prove robust to tackle the problem."
1) "The problem needs to be dynamically complex"
"A first criterion is that the problem needs to be dynamically complex, because of potential underlying feedback processes. In other words, decisions will provoke intended and potentially unintended consequences, which cannot be easily foreseen without the help of a computer model. This excludes all kinds of 'static' problems, i.e. questions which aim at identifying an existing situation at some point in time. In reality, it may be difficult to employ this as a criterion. Let us take the example of a company which goes through a selection process for a new site, or a government which attempts to decide on a site for a waste disposal unit. The question is to determine the best site given a number of criteria to include in the decision process. Typically, a problem like this can be tackled with the aid of techniques from the realm of Multi Attribute Utility Decision making or MAUD (see for instance von Winterfeldt and Edwards (1986)). The process leads to an optimal choice given a number of criteria (e.g. cost, parking space, available room, attractiveness). The same problem however, could also be tackled from the point of view of system dynamics, if one is interested in the potential effects a particular decision might produce. For instance, selecting a particular site will produce all kinds of (un)intended consequences, which could only be established with the aid of a dynamic (computer) model.
"To provide another example, let us take the problem of sickleave in an organization. The question might be what can be done about the high rate of sickleave. This question may lead to two different kinds of approaches. The first is to apply some kind of ideation technique (e.g. brainstorming) in order to generate a large number of potential solutions to the problem. Next, one might select one of these alternatives based on such criteria as implementability, expected success, etc. In other words, the problem is taken for granted and one immediately focuses on potential solutions. Although this method is rather quick and relatively cheap, it entails the danger of fighting symptoms rather than causes. A system dynamics approach on the other hand would focus on the identification of potential causes and feedback loops underlying the problem. Rather than just fighting symptoms a careful analysis might reveal underlying causes which have to be eliminated to solve the problem. The system dynamics model would be used to identify and evaluate robust and effective policies."
2) Typically system dynamics addresses long term problems, frequently those problems which have been tackled unsuccessfully in the past
"Closely related to the first criterion of dynamic complexity is the time horizon. System dynamics models frequently show that a particular policy works well in the short term, but might prove ineffective in the long term and vice versa. Typically, system dynamics addresses long term problems, frequently those problems which have been tackled unsuccessfully in the past. The latter might be an indication that attempted policies have failed, because of systemic compensation. In other words, feedback processes within the system counteract the assumed effectiveness of the policy. System dynamics aims at identifying policies, which prove effective in the longer run. In short, system dynamics is appropriate in situations where (a) the problem is dynamically complex because of underlying feedback processes, and (b) one looks for robust long term solutions."
3) Check whether it is possible to generate a reference mode of behaviour which may represent a problem. If this cannot be accomplished one needs to be particularly careful.
"These directions still leave considerable room and may be difficult to apply in a concrete situation. Fortunately, there is a more tangible guideline to be applied. The best practical approach is to check whether it is possible to generate a reference mode of behaviour which may represent a problem. If this cannot be accomplished one needs to be particularly careful. As an example let us take the following two problem formulations:
1. How can the decreasing trend in the number of ships flying the Dutch flag be reversed?
2. How could this organization best be designed?
In the first case, I am easily able to produce a reference mode of behaviour, while in the second this is virtually impossible. Problems like the second should thus be treated with caution when system dynamics is concerned.
"As pointed out in Chapter 2, there does not necessarily need to be an empirical (historical) reference mode of behaviour. A hypothetical reference mode may well suffice. In Chapter 7 for instance, I will present the problem of the Dutch housing corporations. As explained in Chapter 2, their primary objective is to build sufficient low cost housing (i.e. the social goal). Until the beginning of the 1990s these corporations did not have to worry about money, they were heavily subsidized by the Dutch government. Recently, however, the government decided that housing corporations would have to operate on the open market. This created a completely new situation, in which the corporation's survival was no longer guaranteed by government support. The question then was whether, and under what conditions, the social goal could be accomplished, while simultaneously ensuring financial continuity. In this case there was no real reference mode of behaviour to guide the model building process, but still a system dynamics approach proved to be useful, because one might construct a hypothetical reference mode in the form of a couple of time series. One of these time series would represent the social goal (will we build enough low cost housing) while the other would represent the financial continuity. Typically, time series like these can be generated by a system dynamics model, which may be an indication that system dynamics will be an appropriate method. Stated somewhat bluntly, a system dynamics model will produce time series for variables in the model. If one gathers that these time series will be able to provide answers to the problem, then system is a likely candidate to tackle the problem. Take the example from the project for the Royal Dutch Medical Association (RDMA), mentioned in Chapter 1. One of their questions was related to the decreasing membership level. This question may typically point towards system dynamics. However, another question was related to the status of the so-called 'scientific associations' with the RDMA namely, how to best incorporate these into the structure of the RDMA. The latter question generally does not point towards system dynamics as a problem solving methodology, it is related to such subjects as legal rights....etc."
4) Check whether the problem description gives rise to thinking in terms of flow processes
"Another practical guideline is to check whether the problem description gives rise to thinking in terms of flow processes. In Chapter 1 I introduced the case of the health care insurance organization that was looking for ways to decrease health care costs. Basically, what one attempted to accomplish was a change in the flow from patients at the family doctor to the medical specialist. This is also an indication that system dynamics might be appropriate."
Vennix, Jac A. M. (1996) Group Model Building: Facilitating Team Learning Using System Dynamics. Chichester, England: John Wiley and Sons
Winterfeldt, D. von and Edwards, W. (1986) Decision Analysis and Behavioral Research, Cambridge, Cambridge University Press