Strong uncertainty

Ignorance and indeterminacy

A repeatedly stated aim of the scientific community addressing the enhanced Greenhouse Effect is to reduce, evaluate and quantify uncertainties. This is reflected in calls for risk—benefit analysis and similar deterministic quantitative approaches. However, contradictory messages appear across reports and within single documents. Such approaches sit uneasily with the recognition that: 'Because of the uncertainties associated with regional projections, of climate change, the report necessarily takes the approach of assessing sensitivities and vulnerabilities of each region, rather than attempting to provide quantitative predictions of the impacts of climate change' (Watson et al., 1997: vii). The scenarios which informed the IPCC work under the TAR were explicitly stated to be 'equally valid with no assigned probabilities of occurrence' (Nakicenovic et al., 2000: 4). However, the TAR then used a classification of 'subjective' probabilities with precisely defined quantitative confidence levels (see IPCC Working Group I, 2001: 2). There is an obvious discrepancy between the different approaches to uncertainty, resulting in inconsistent statements.

In the last chapter the way in which uncertainty pervades the issue of predicting climate change was explored from the perspective of weak uncertainty. That approach concentrates upon a specific characterisation of missing knowledge as 'objective risk'. The idea of strong uncertainty was also introduced and discussed as being relevant to both scientific and economic modelling and prediction. The aim in the current chapter is to further elucidate the distinction between different types of uncertainty, their relevance to climate change and implications for research and policy. This requires a more in-depth explanation of the classification already introduced before moving on to the issues raised by strong uncertainty.

Risk can be defined as the case where the set of all future events are known but the occurrence of any one event is only a potential. Thus, tossing a fair (evenly weighted) coin would be regarded as leading to two events, either heads or tails, with a known probability of 50:50. In common use risk is associated with harm or negative outcomes, but under environmental risk assessment or economic decision analysis the outcomes could all be beneficial, neutral or harmful or any mixture of the three.

The standard theory of decision-making under uncertainty modifies the normal economic assumption of perfect knowledge to account for risk. That is, rather than being certain as to the outcome of their choices, the decision-maker (e.g. consumer, producer, civil servant or politician) still knows all future possible outcomes but only the probability (density function or distribution) of their occurrence. This allows the calculation of every possible outcome and the expected value (or utility) of each choice, also called the probability weighted average. Additional weightings can be added to account for the attitude of the decision-maker towards risk, i.e. risk neutral, risk averse or risk taker. The original economic decision model can be maintained by using the expected values instead of the known ones. This theory assumes that 'objective' probabilities are associated with events or outcomes.

Empirically observable and repeatable events, such as a coin toss, allow the construction of what are termed 'objective' probabilities or risks of an event occurring. That is, anybody could repeatedly toss the coin and count the number of heads and tails and would find the same probability of their occurrence. However, as Loasby (1976: 8) states, 'the notion of an objective probability distribution carries a strong (but unstated) implication about the nature of the world, namely that it generates all the necessary (and quite unambiguous) frequency distributions from a stable population of events'. As already shown in chapter 4, this is not the case for global climate change and, as Loasby notes, is indeed a generally implausible requirement. As also explained in the last chapter, the past provides no assurance of the future.

Accepting that 'objective' probabilities are absent means moving to what economists term 'uncertainty', which is in fact still only a limited, although problematic, adjustment. In the absence of an ability to estimate a probability distribution from observation (i.e. 'objective' probabilities) an appeal may be made to 'subjective' probabilities as an alternative. The assumption is now that probability functions are undefined although all states of the world or future outcomes are still known. The solution to this problem for mainstream economists is to allow the use of probabilities placed by the decision-makers themselves upon the likelihood of an event. In practice various methods are employed to obtain such 'subjective' probabilities. This new adjustment makes the move from 'objective risk' to 'subjective risk', but neither has addressed what is commonly understood as uncertainty. Thus, economists (and scientists) tend to restrict themselves to discussions of weak uncertainty.

Keynes (1988), amongst others, argued that the terms risk and uncertainty should be regarded as strictly separate.1 However, in common use the terms are often interchanged so that employing the term 'uncertainty' in such an unusually restrictive way (i.e. excluding risk) can create confusion; the temptation is always to slip back into common usage. Therefore, here, risk is included under weak uncertainty while the term strong uncertainty is applied to the more Keynesian concepts, as shown in table 5.1.

122 Strong uncertainty: ignorance and indeterminacy Table 5.1 Classifications of uncertainty, risk and ignorance

Sub-categories Explanation

Weak uncertainty Objective risk

Subjective risk

Strong uncertainty Partial ignorance Indeterminacy

Known outcomes and their probabilities; termed 'risk' by mainstream economists; probabilities given by natural world Known outcomes only; termed 'uncertainty' by mainstream economists; probabilities assessed from human preferences Unknown outcomes Unpredictable outcomes

Strong uncertainty then refers to the admission of a lack of knowledge about potential outcomes. The coin may land on its edge or disappear between the floor boards. Such strong uncertainties are often excluded from calculation because they are regarded as so unlikely as to be of negligible significance, in which case, we are truly surprised when they occur. They often relate to events which have been excluded by assumption. Thus, partial ignorance is an inevitable part of modelling where situations are simplified and vision restricted in order to aid understanding. Strong uncertainty requires that allowing for surprise events and admitting knowledge about future possible events is always incomplete. This is particularly relevant to complex systems (such as those forming climate, or economies) where choice cannot be assumed to be fully informed (contrary to the simple coin toss, and assumptions of positive economics).

As Loasby (1976) has explained, choice within complex systems provides the basic subject matter of economics. Thus, what he terms 'partial ignorance' becomes central to economics because the subject concerns the study of the unintended social repercussions of human actions. As a result the problem of how to describe and deal with a lack of knowledge in the sense of ignorance opens up a much wider debate. The concept of ignorance raises the idea of an irreducible lack of knowledge which is never removed by research and is in fact endemic to scientific knowledge. As a result ignorance is revealed due to events external to an individual's or group's models, disciplinary focus or world view and this can force a changed of perspective.

In addition, the concept of an 'indeterminacy' of outcomes is also relevant. While the concept of an indeterminate future may appear similar to a state of partial ignorance there are additional distinct features. Indeterminacy will arise due to a lack of knowledge and an inability to comprehend all existing knowledge, but also because of the social context within which knowledge is applied. Social context varies and involves customs, culture, institutions and socio-economic systems. The importance of indeterminacy is explained later in this chapter.

First in the following section five events are used to characterise the range and type of uncertainty confronting society due to the enhanced Greenhouse Effect. This draws upon chapter 3 but aims to focus the reader's mind on how some of the issues have been debated, judged or neglected. The way in which weak uncertainty has been used to address such issues is then discussed with specific focus on the insurance risk model. This shows flaws in the economic approach which are then explained in greater detail. The need to move away from the weak uncertainty characterisation is explained and the alternative explanation using strong uncertainty described. The current conception of scientific and technological research as removed from social processes is highlighted as false. Knowledge about natural and socioeconomic systems is then placed within a common context of choice where humans simplify, assume and guess.

Was this article helpful?

0 0

Post a comment