An essential aspect of strategic planning is the high level of uncertainty. How do we deal with this? The general answer to this question would be to make robust and flexible decisions. Every planning process should be sustainable, even when conditions change, in order to achieve optimum (business) results in every individual case. Read more on this in our blog: Planning under uncertainty – How does it work? Here, we want to take a look at scenario generation – in other words, how to achieve optimum planning.
Shaping the future with scenarios?
So how do we make a robust and flexible decision which can also apply to the future? In general, it’s hard to imagine a reality which doesn’t even exist yet. Our powers of imagination come to a halt at the following hurdle: There is not only one, but a number of possible futures. The world can keep turning in so many different ways. At some point, man’s imagination can go no further and this is where scenario generation comes in to create an optimum planning process under uncertainty.
Scenarios as a basis for a “good tomorrow!”
Scenarios are used to actively prepare for the future and also to highlight uncertainties in planning. However, they aren’t fantastical ideas which reflect the wishes and fears of managers and planners.
Scenarios are draft versions of possible future developments – for instance in demand, market and price developments, costs… They are also based on the current situation as well as on trends and changes we expect in the future. They are not an end in itself. Measures and options can be derived from these scenarios, helping businesses to decide what action to take, in production planning for example, and they can define them for every scenario.
As a rule of thumb: a scenario comprises only the aspects that a company cannot exert influence upon.
However, scenarios are also useful when reviewing current business plans and strategies. The aim is test their robustness or detect any possible weaknesses and make the necessary changes with a view to achieving the optimum result. Read more on robustness and flexibility in our blog: Planning under uncertainty – how does it work?
Scenario generation
The technological basis for planning under uncertainty is the use of suitable scenario generators. However, there are strict quality requirements for the scenarios that these generate because they provide the basis of optimum planning.
Scenarios should meet the following three requirements:
- consistency
- completeness
- controllability
Consistency
Every scenario should be intrinsically consistent, in other words, conflict-free. For example, it may be that increased price elasticity combined with a decreased market demand is highly unlikely if they are both closely related to economic growth.
Furthermore, we need to bear in mind that the uncertainty typically increases with the long-term nature of the prognosis. This aspect should also be recognizable in the scenarios.
Completeness
Ideally, the scenarios should completely cover the period of possible future developments. This includes not only extremum but also variability over time.
Diagram 1 is an example of a basic scenario generator. Here, the variability, e.g. the demand/market development over time has not been considered.
In comparison, diagram 2 shows scenarios with a high variance over time.
Diagram 1 – Example of a basic scenario generator without variability over time
Diagram 2 – Example of a scenario generator with variability over time
The diagrams show that far less flexibility is required for scenarios created by a basic scenario generator which also consider variability over time. This is why it is essential to really consider all possible future developments when it comes to the planning strategy.
Controllability
The number of scenarios must be limited to achieve a fast enough calculating time. In addition, the number of scenarios should not increase too much if further aspects of uncertainty such as exchange rates or additional markets are also taken into consideration.
Basic scenario generators first generate individual part-scenarios for every uncertainty and then combine each one with the other part-scenarios to create a complete scenario. In five market regions with the demand characteristics “low”, “normal” and “high” this leads to a fairly transparent 243 scenarios. However, if we look at the market more closely and take 10 regions each with five characteristics, you get almost 10 million scenarios.
In addition to this, the aspect of variability over time has not been taken into consideration – the period of possible developments has then been covered completely, even with almost 10 million scenarios.
It would seem that the demand for completeness is in conflict with controllability. How can we achieve both?
There are two approaches here that can be combined with one another:
- At first, a large number of scenarios are generated and these are then reduced, for instance with the help of DOE (Design of experiments)
and/or
- The scenarios are specifically generated so that they already show a comparatively low number of good features.
The important thing is to create consistent scenarios which cover the period of possible developments completely yet at the same time are controllable in number.
OPTANO and scenario generation
A scenario generator has been developed and implemented for OPTANO which has created an ideal balance among all the requirements mentioned and provides a suitable choice of scenarios as a result.
The scenarios generated by the scenario generator are the basis for the further procedure of planning under uncertainty. The aim is to support the user in developing suitable solutions and strategies, e.g. in investment or production planning. One possible method here is stochastic optimization – a specialised procedure for modeling and solving optimization problems under uncertainty.
Read more about stochastic optimization – a further step towards optimum results for planning under uncertainty, coming up in our next OPTANO blog.
Author: Alexandra Navarra
Other topics you may find interesting….
Book recommendation: Hello World: How to be Human in the Age of the Machine
Book recommendation: Humble Pi: When Math goes Wrong in the Real World.
powered by Borlabs Cookie 
