Modelling process are often described in the context of helping to make a “decision”. This can be a useful way to think about models, as it helps to frame the situation, define the objectives, a determine many other modelling requirements. (However, it should also be interpreted in a broad sense: Decisions are not only of the “go/no go” variety, and models can be used to design a range of possible actions and assess the implications of each within different future scenarios, and so on.
The following provides an initial description of the link between decision-making and the role of modelling.
Let us assume that someone is faced with a decision as to whether to go on vacation or not:
Of course, there could be many factors which influence the choice between the decision options. At the simplest level, it could be that going on vacation is not at all a valid option, since one not be able to free up the necessary time, or one may suffer from health issues which limit one’s travel possibilities.
However, if going on vacation is potentially viable (as is not going), then it would be helpful to identify the criteria on which the choice between them is to be based. For example, the decision to go (or not) may be driven by budget considerations only: One will go on vacation if the level of savings that one will have next year is sufficient:
Once again, in some cases it may be immediately clear that one or other option should be pursued (or not). For example, one may already have more than enough savings to pay for the vacation in any case, or alternatively one may be currently paying to renovate one’s apartment, and already know that there will be no spare budget.
On the other hand, there are cases where the decision is not so clear, and one needs to do a more analysis of the savings that one will have in the future before a decision can be made. The analysis can also help to “optimize” one’s behaviour. For example, if the level of savings will be sufficiently high, then some of the excess money could be used for other purposes (such as investments in higher return projects), rather than being held in an instant access (low interest bearing account).
Then, the role of “modelling” would be to seek to understand one’s future (or total) savings position in detail (i.e. based on capturing the elements which determine this and expressing the relationships between them). For example, a simple model could be used to forecast one’s future savings based on one’s income, expenditure, and current savings and so on:
The outputs of the model are therefore the quantification of the decision criteria.
In summary, starting with the overall context, we have identified the decision possibilities, and the criteria, and then built a model to evaluate these:
Note that it can be considered that the model is designed “backwards” (i.e. starting with the context, decision and criteria), but is built and calculates “forward”. It would be possible to continue the process, and to build a more detailed model. For example, one could break down living expenses into its sub-components, or consider various scenarios for future salary development, and so on.
The modelling process can therefore help one to understand the factors that can be influence in order to change the level of savings, or to understand the possible scenarios and risks involved, as well as to optimize one’s overall behaviour in this situation.
It may be that some decision criteria cannot easily be built into a model. For example, a detailed reflection of the situation may lead one to be concerned about the environmental impact of taking a flight, or of the political context in the country that one would visit:
Of course, in an ideal world, one would find a way to evaluate all decision criteria within a model. However, in practice this may not be possible: In general, the decision-maker will then need to weigh up the various criteria. This may involve making intuitive or judgmental trade-offs, which may involve less transparency and be subject to personal or group biases.
Thus, an important role of modelling is to bring as much transparency and objectivity into the process as is possible or appropriate, even if this may not always be able to be done completely.