Defining an Uncertainty

Notes on “Defining Uncertainty: A Conceptual Basis for Uncertainty Management in Model based Decision Support” Integrated Assessment 2003, Vol 4, No 1 pp 5-17 Walker et al.

1. Uncertainty is not just the absence of information; it is both the lack of applicable knowledge (due to inexact, unreliable or relative ignorance of the information available) and the inherently variable nature of people and the situation that surrounds them.

2. Uncertainty depends on the perspective of the person considering the uncertainty.

   1. Planners focus on the accumulated uncertainties associated with a projected outcome and how to make the plan as robust as possible to achieve that outcome.

   2. Decision-makers focus on how uncertainty increases or decreases the value of an outcome in terms of achieving their possibly conflicting goals, objectives and interests.

   3. The ultimate goal of decision making in the face of uncertainty should be to reduce the undesired impact of surprises, not to eliminate them completely.

   4. This paper is from the perspective of the Planner.

3. There are three dimensions of uncertainty: The location of the uncertainty, the level of the uncertainty and the nature of the uncertainty.

4. The location of the uncertainty is where the uncertainty exists within the whole complex model, so the location varies based on the description of the plan. The generic locations include:

   1. Context: Whether the uncertainty is located within the boundaries of the plan or outside the boundaries of the plan.

   2. Plan Uncertainty: The conceptual plan has two types of possible uncertainty

      1. The plan structure: that is uncertainty about the nature of the plan itself, and

      2. The technical structure: that is the uncertainty about the computer implementation of the plan to create a model.

   3. Inputs: Plans depend on both the description of the current system and a description of the outside forces that are driving change in the current system. Therefore input uncertainty can be both about the parameters of the data inputted to the plan, and the description of the impact of outside drivers of change will have on the decision-makers goals.

5. Levels of Uncertainty: The level of uncertainty determines how robust and adaptable the plan should be to reduce the negative impact of surprises on achieving the decision maker’s goal. The levels of uncertainty are determinism, statistical uncertainty, scenario uncertainty, recognized ignorance and total ignorance.

   1. Statistical Uncertainty: Some uncertainty can be adequately described in statistical terms, in terms of deviation from a norm. An example of statistical uncertainty is when there is imprecise sampling, or measuring of the sample, upon which the statistical norm and deviation is based. There is, however, relative certainty about what the model is accurate and fully describes the relevant driving forces and outcomes over a short period of time.

   2. Scenario Uncertainty: Uncertainty related to the external environment of a system, which is internally plausible and consistent, is described in scenarios. Scenario uncertainty implies that there is a range of outcomes of the analysis, there is uncertainty about which driving forces are relevant to the outcomes desired, and uncertainty about the levels of the relevant changes over a longer period of time.

   3. Recognized Ignorance: This is the fundamental weakness understanding the functional relationships, the statistical properties and the external driving forces in the system. If ignorance is recognized, some can be reduced through research and development, but some ignorance always exists and cannot be reduced to zero.

6. The Nature of Uncertainty: The nature of our uncertainty can be both to our imperfection of knowledge and due the inherent variability of human and natural systems concerning individual, social, economic and technological developments.

   1. Knowledge: Limited or inaccurate data, measurement error, limited understanding, imperfect models, subjective judgments, ambiguities and so forth create uncertainty in the nature of knowledge in a specific plan. This can be reduced by better research and data collection, but never eliminated.

   2. Variability: The real world varies over space and time, due to the inherent randomness of nature, human nature, social, economic and cultural dynamics as well as technological surprise.

   3. A common mistake occurs when there is no distinction between uncertainty inherent in variability uncertainty (which cannot be reduced) and knowledge uncertainty (which can be reduced).

Examples of Uncertainties as statements

• Over the next ten years, we will either have hyper-inflation or persistent deflation in the global economy.

• Over the next year the US Dollar will either weaken or strengthen against a basket of other currencies.

• In the next five years a member of the next generation of the family either will receive their MBA or they will not.

• Over the next ten years, commercial bank loans will either become much easier or much harder to get.

• During the next ten years, I will either die (and estate taxes will need to be paid) or I will not die, and no estate taxes will need to be paid.

Let us say that one of the critical uncertainties is determined to be that, over the next ten years, the global economy will either have hyper-inflation or persistent deflation. Now we have to define this uncertainty, and what are the primary and the secondary characteristics of this uncertainty. This is the prerequisite to developing planning models, or scenarios, that are both robust and adaptable.

Applying the Matrix

Uncertainty Parameters:Over the next ten years, the global economy will either have hyper-inflation or persistent deflation

The primary characteristics (and where we should focus time and money on planning) of this uncertainty are:

1. The range of economic and financial external factors affecting the outcome.

2. The range of social factors that affect the outcome

3. The range of political and regulatory factors affecting the outcome.

4. The range of innovation factors affecting the outcome.

5. The range of wealth management factors affecting the outcome.

6. The range of Competiveness factors affecting the outcome.

The Secondary elements of this uncertainty are:

1. Global resources will be predictable as a deviation from a norm.

2. Education and learning will have a predictable affect as deviation from a norm

3. Environmental change will have a predictable affect as deviation from a norm

4. Personal change will have a predictable affect as deviation from a norm

5. Systems Data will be predictable as a deviation from a norm

The nature of the uncertainty is:

2. Global Resources knowledge limitations and inaccuracy will affect the outcome

3. Global Resources will have some random variability effect

4. The structural soundness of the models have knowledge limitations and inaccurate data which will affect the outcome

5. The mathematical techniques of the models have knowledge limitations and inaccurate data which will affect the outcome