5. Burocracia y represión: obstáculos para la democratización
5.1. Obstáculos burocráticos en el desarrollo de las mesas e implementación del AF
InChapters 5–7 we have seen how risk assessments and risk management are influenced by the risk perspectives. Now we would like to go one step further, to provide guidance on what should be the preferred approach to risk. The basis for the guidance is the discussions in the previous chapters. Firstly we need to clarify what we mean by risk. A number of definitions and interpret-ations of the risk concept exist as discussed inChapter 2. Many of these are probability-based. Below (Section 8.1) we present and discuss a structure for characterising the definitions, which is founded on a clear distinction between (Aven,2010f)
(a) risk as a concept based on events, consequences and uncertainties;
(b) risk as a modelled, quantitative concept; and (c) risk descriptions.
The discussion leads to an approach for conceptualising and assessing risk, which is based on risk defined by (a), i.e. is founded on the (A,C,U) risk perspective, and the probability-based definitions of risk can be viewed as model parameters and/or risk descriptions. The approach provides clear guidance on how to think when conceptualising and assessing risk in practice.
Next in this chapter (Section 8.2) we present and discuss a general model-based framework for risk assessments. Starting from an industry guide to quantitative uncertainty analysis and management, clarifications and simpli-fications are made to ensure consistency with the (A,C,U) risk perspective.
Some simple examples are included to motivate and explain the basic ideas of the framework.
In risk assessments, probability is the common tool used to describe the epistemic uncertainties about unknown quantities. However, the purely probability-based approaches to risk and uncertainty analysis can be challenged as we have discussed throughout this book. A key point is that the support of 138
the probabilities is not reflected by the numbers produced. This concern has sparked a number of investigations in the field of uncertainty representation and analysis, which has led to the development of several alternative approaches, including possibility theory and evidence theory. These theories and methods represent strong research areas and in the last section of this chapter (Section 8.3) we question to what extent the raised challenges of the probability-based methods can be solved by these approaches.
8.1 What is risk? A structure for conceptualising and describing risk Risk is a fundamental concept for most scientific disciplines, but no consen-sus exists on how to define and interpret risk. Some definitions are based on probabilities, some on expected values, and others on uncertainty. Some consider risk as subjective and epistemic, dependent on the available know-ledge, whereas others grant risk an ontological status independent of the assessors. The situation is chaotic and leads to poor communication. We are also afraid that it hampers effective risk management as well as the development of the risk field, as many of these definitions and interpretations lack proper scientific support and justification.
Of course, business needs a different set of risk methods, procedures and models from, for example, medicine and engineering. But there is no reason why these areas should have completely different perspectives on how to think when approaching risk and uncertainty, when the basic challenge is the same – to conceptualise that the future performance of a system or an activity could lead to outcomes different from those expected, desired, planned, or not in line with stated objectives.
Think of an activity in the future, say the operation of an offshore instal-lation for oil and gas processing. We all agree that there is some risk associ-ated with this operation. For example, fires and explosions could occur leading to fatalities, oil spills, economic loss, etc. But it is not straightforward to explain what we mean by this risk if we require a precise definition and would like to use the concept in scientific studies. Risk analysts would introduce a set-up which directly or indirectly defines how risk is understood and assessed; refer to Chapters 5 and 6. The set-up would typically be probability-based, with probabilities interpreted either as relative frequencies or as subjective probabilities. All such set-ups can be challenged as not being able to reflect risk in a proper way. Important risk aspects could be camou-flaged or hidden by the set-up. Discussions of the set-up are therefore import-ant, not only from a theoretical point of view but also from a practical risk management perspective.
8.1 The meaning of risk 139
We have identified several definitions of risk that can be used as an overall, common definition. They all belong to the category (a). Many attempts have been made to establish a unified risk perspective, but none of these have obtained broad acceptance in practice. There could be many reasons for this.
Firstly, the scientific work on risk may not have reached a sufficiently mature level for establishing such a definition. The exploring phase is not completed.
Secondly, the scientific literature has a focus on the generation of new ideas and suggestions, and on a critique of other contributions. By its nature, it is hard to obtain broad consensus on scientific issues in general and risk definitions in particular. And thirdly, the standardisation organisations have not been able to produce sufficient broad and precise definitions which could be accepted by the scientific expertise on risk.
Consider for example the latest definition from the International Stand-ardisation Organisation (ISO, 2009a,b): risk is the effect of uncertainty on objectives. What does this mean? Risk has to do with uncertainty, but is it the effectof uncertainty? And risk is related to objectives, but what if objectives are not defined? Then we have no risk? Asking experts on risk, there is no doubt that this definition would lead to numerous different interpretations.
The definition is not sufficiently precise, and one may certainly also question its rationale as indicated.
In Chapter 2 we presented and discussed a set of common definitions of risk, including (the numbers 1–8 are the same as those used inChapter 2) 0. Risk equals the expected loss (Verma and Verter,2007; Willis,2007).
1. Risk is a measure of the probability and severity of adverse effects (Lowrance,1976).
2. Risk is the combination of probability and extent of consequences (Ale,2002).
3. Risk is equal to the triplet (si, pi, ci), where si is the ith scenario, piis the probability of that scenario, and ci is the consequence of the ith scenario, i¼ 1,2, . . . N (Kaplan and Garrick,1981).
4. Risk refers to uncertainty of outcome, of actions and events (Cabinet Office,2002).
5. Risk is a situation or event where something of human value (including humans themselves) is at stake and where the outcome is uncertain (Rosa,1998,2003).
6. Risk is an uncertain consequence of an event or an activity with respect to something that humans value (IRGC,2005).
7. Risk is equal to the two-dimensional combination of events/consequences and associated uncertainties (Aven,2007a,2010e).
8. Risk is uncertainty about and severity of the consequences (or outcomes) of an activity with respect to something that humans value (Aven and Renn, 2009a).
For the measures that are based on probabilities and expected values, we may generate two versions, one where the probabilities are interpreted as relative frequencies (and the expected values as averages), and one where the prob-abilities are subjective (knowledge-based) probprob-abilities (and the expected value is interpreted as the centre of gravity of the probability distribution).
We write definitions xf and xs, respectively, to separate the two categories, x¼ 0, 1, 2, 3. Consider as an example category 0, risk defined as the expected loss. According to definition 0f, risk is understood as the average loss when considering an infinite number of similar situations, whereas 0s means that risk is the centre of gravity of the subjective probability distribution of the loss. Following the suggested structure for characterising the various risk definitions we have to place these definitions in one of the categories (a), (b) (c), defined above.
The result is that definition 0fis in category (b) and 0sis in category (c), as risk in the former case is based on the model of an infinite number of similar situations and risk in the latter case is a way for the assessor to describe or characterise risk. The expected loss Eswhen using subjective probabilities is a risk index based on the background knowledge (K) of the assessor. A similar analysis is carried out for the other eight definitions. The result is shown in Table 8.1.
We refer toChapter 2for a discussion of these and other risk definitions.
Table 8.1 Categorisation of the nine risk definitions according to the structure (a)–(c)
Risk definition Category
0f b
0s c
1f b
1s c
2f b
2s c
3f b
3s c
4 a
5 a
6 a
7 a
8 a
8.1 The meaning of risk 141
If relative frequency-interpreted probabilities Pfconstitute the basis (defin-itions 0f, 1f, 2f, and 3f) risk is a modelled, quantitative concept (category b) and we may formalise the definitions by writing
Risk ¼ ðA; C; PfÞ;
where A represents the events (initiating events, scenarios) and C the consequences of A, as inSection 2.2.
If, on the other hand, subjective (knowledge-based) probabilities constitute the basis (definitions 0s, 1s, 2s, and 3s), the definitions must be viewed as risk descriptions as they express the analysts’ (experts’) degree of belief concerning A and C. Also the background knowledge K that the probabilities are based on should be considered a part of the risk description.
If we search for a widespread agreement on one definition of risk we have to look among the categories (a). The others have to be excluded as they are based on either a model or an assignment of uncertainty using the tool, subjective (knowledge-based) probability. Risk should also exist as a concept without mod-elling and subjective probability assignments. We face risk when we drive a car or run a business, also when probabilities are not introduced. For risk assessment we need the probabilities, but not as a general concept of risk. In this way we obtain a sharp distinction between risk as a concept and risk descriptions (assessments).
As discussed inSection 2.5, definition 4 (which basically says that “risk¼ uncertainty”) cannot be used as it fails to include the consequence dimension.
Hence we are led to two candidates among the a-definitions: the (A,C,U) definitions (7–8) and the (A,C) definitions (4–6). The latter group means that the common risk terminology has to be revamped (refer to discussion in Section 2.5) and we therefore prefer to use the (A,C,U) definition.
Risk is thus defined. The next stage would then be to specify how to describe risk. We seek a general structure and we cannot base it on the use of frequentist probabilities (chances) as these cannot be meaningfully defined in all cases.
However, knowledge-based probabilities can always be defined, and they are introduced as the recommended tool for describing the uncertainties.
This leads to a risk description as was first noted inSection 2.8:
Risk description¼ ðA; C; U; P; KÞ;
that is, risk is described by events A and consequences C, subjective (knowledge-based) probabilities P, uncertainties U not captured by P, and K the background knowledge that U and P are based on. The U component may for example be a qualitative assessment of uncertainty factors (assump-tions that the probabilities are based). A subjective probability P(A) ¼ P(A | K) is interpreted as a knowledge-based probability with reference to
an uncertainty standard expressing the assessor’s uncertainty about the occurrence of the event A given the background knowledge K. Following this interpretation the assessor compares his/her uncertainty (degree of belief) about the occurrence of the event A with the standard of drawing at random a favourable ball from an urn that contains P(A) · 100%
favourable balls (Lindley, 2000).
However, also in this setting we may establish relative frequencies, but they are referred to as chances and not probabilities. A chance is the limit of a frequency of similar (formally exchangeable) random events. More generally we introduce probability models with unknown parameters. A chance is an example of such a parameter. By the Bayesian updating machinery, know-ledge about the parameters is described first by the prior distribution, then updated to produce the posterior distribution to reflect observations. Finally, this distribution is used to generate the predictive distribution of the events A and consequences C. These predictive distributions then incorporate the variation reflected by the probability model (and the chances) and the epi-stemic uncertainties about the true value of the parameters. The main features of the thinking are shown inFigure 8.1. Models will be further discussed in the coming section.
Note that chances and probability models are tools used to describe risk.
They are not identified as risk per se. This is in contrast to the “Risk¼ (A,C,Pf)”
types of approaches, including the probability of frequency approach (see Section 2.5), where the relative frequency-interpreted probabilities (chances) Pf always need to be defined. They constitute the foundation of the approach. In the (A,C,U) types of approaches, chances are only defined when exchangeable sequences can be justified. Chances need some sort of
Activity Risk
(A,C,U)
Risk description (A,C,U,P,K), based on – Knowledge about phenomena
– Models (including probability models, chances) – …
Figure 8.1 The main elements of the recommended risk approach.
8.1 The meaning of risk 143
model stability (Bergman, 2009): populations of similar units need to be constructed (formally an infinite set of exchangeable random variables). We will, for example, not define a chance p of a terrorist attack (Aven and Renn,2009b); it has no meaning, as also mentioned inSection 7.3.
It may be a challenge to reveal and describe all the uncertainties. Qualita-tive approaches can be used as indicated inChapter 6. See also the discussion inSection 8.3.
8.2 A model-based framework for risk assessments
A guide to uncertainty analysis and management in industry has recently been issued (de Rocquigny et al., 2008). The guide is written by a project group of the European Safety, Reliability and Data Association (ESReDA).
The book project was motivated by the fact that no authoritative standard exists for how to analyse and quantify uncertainty. The guide presents a numbers of practical cases, all based on the same uncertainty analysis frame-work; seeFigure 8.2. As uncertainty is a main component of risk as defined in the previous section, this framework for uncertainty assessment is highly relevant to risk assessments. The discussion in this section is to a large extent based on Aven (2009c,2010b).
The key variables of interest are denoted Z (which could be a vector). To assess Z a model G(X,d) is introduced which links a set of input variables X and some fixed quantities d to Z (also X and d could be vectors). To describe the uncertainties, probabilistic and non-probabilistic methods (for
G(X,d)
Sensitivity analysis and importance ranking
Decision criteria P < p0 Feedback
process
Uncertainty propagation
Figure 8.2 The overall framework adopted by the uncertainty analysis guide (de Rocquigny et al.,2008).
example possibility theory and evidence theory, seeSection 8.3) can be used.
A common approach is to use a parametric probability distribution (wherem is the parameter) to establish a probability distribution for X. Using the model G, an uncertainty description is obtained for Z. The tool used for this purpose could be an analytical approach or Monte Carlo simulation. Some quantities of interest, for example expected values and variances, are specified and computed from the measure of uncertainty derived, i.e. the probability distribution of Z. These quantities provide input to a decision process, which could be based on some decision criteria expressing for example that a probability should not exceed a specified level. Sensitivity analysis provides insights about how the input quantities affect the output quantities, and importance ranking identifies what factors, subsystems etc. are most import-ant based on some defined criteria, for example the contribution to the variance. The result of the analysis may lead to some action (feedback process), for example that there is a need for design changes to meet the criteria. The actions need to be seen in relation to the goals of the analysis which usually fall into the following categories (de Rocquigny et al.,2008):
Understand: To understand the influence or rank the importance of uncer-tainties, and thereby to guide any additional measurement, modelling or research and development efforts.
Accredit: To give credit to a model or a method of measurement, i.e. to reach an acceptable quality level for its use. This may involve calibrating sensors, estimating the parameters of the model inputs, simplifying the system model physics or structure, fixing some model inputs, and finally validating according to a context-dependent level.
Select: To compare relative performance and optimise the choice of main-tenance policy, operation or design of the system.
Comply: To demonstrate compliance of the system with an explicit criter-ion or regulatory threshold.
Most analysts and researchers would probably consider this framework a logical and useful structure for performing uncertainty analysis in practice.
There is not much that is controversial or problematic about the framework described at this overall level. However, when we go into the details, the meaning and use of the different concepts are not so straightforward as we will see from the coming analysis.
Risk analysis may be considered more restricted than uncertainty analysis as risk analysis focuses on future events, whereas uncertainty analysis is concerned with uncertain quantities, whether they relate to the future or not. However, the frameworks and tools used for analysing risk are to a large extent general and in most cases they are applicable also for “non-future”
8.2 A model-based framework for risk assessments 145
type of situations. Anyhow, uncertainty is the key concept to be addressed and we need to clarify:
(i) What are the uncertain quantities?
(ii) Who is uncertain?
(iii) How should we represent the uncertainties?
In the framework illustrated by Figure 8.1 the uncertain quantities are X and Z, but in practice it may not be straightforward to choose the appro-priate X and Z as we have seen from the discussions inChapters 5and6. Are X and Z observable quantities like time to failures and costs, or parameters of probability models? Uncertainty about the quantities X and Z raises the question: who is uncertain? Is it the decision-maker, the analyst or some experts used in the assessment? We will argue that the uncertainty is normally that of the analyst. Experts and others can produce input to the analyst but the analyst has the ownership of the final distributions and quantities of interest. Of course, in some cases the aim of the analysis is simply to report the knowledge expressed by some experts, but as the analysts are responsible for how to elicit this knowledge and analyse it, care should be shown in presenting the results as independent of the analysts. Being precise on the ownership is essential for obtaining a clear understanding of the framework and how to communicate its results. For a further discussion of this issue, see Aven and Guikema (2010).
To express the uncertainties an adequate representation is required, and probability is the natural choice as it meets some basic requirements for such a representation (Bedford and Cooke,2001, p. 20):
Axioms: Specifying the formal properties of the uncertainty representation.
Interpretations: Connecting the primitive terms in the axioms with observ-able phenomena.
Measurement procedures: Providing, together with supplementary assump-tions, practical methods for interpreting the axiom system.
Many types of uncertainty representations exist, but many fail when it comes to interpretation. We should reject a representation which has no clear interpretation. It is not sufficient to say that a measure expresses for example a degree of belief. We need to know what it means that the measure is 0.2 instead of 0.4.
The present analysis has a focus on the use of probability to measure uncertainty, although the de Rocquigny et al. (2008) framework allows for both probabilistic and “non-probabilistic” representations of uncertainty. We refer to the discussion inSection 8.3.
8.2.1 A modified framework
It is possible to simplify ideas and clarify key concepts in the de Rocquigny et al. (2008) framework when restricting attention to probabilities as a meas-ure of uncertainty. SeeFigure 8.3. The basic features of the modified
It is possible to simplify ideas and clarify key concepts in the de Rocquigny et al. (2008) framework when restricting attention to probabilities as a meas-ure of uncertainty. SeeFigure 8.3. The basic features of the modified