2.1 Main Characteristics of Risk Evaluation
The two central categories of risk evaluation are the extent of damage and the probability of occurrence (for definitions see Knight, 1921; National Research Council, 1983; Fischhoff ct al„ 1984; Fritzsche, 1986; Short, 1984; Bechmann, 1990; IEC, 1993; Kolluru and Brooks, 1995; Banse, 1996; Rosa, 1997). Damage should generally be understood as negatively evaluated consequences of human activities (e.g., accidents by driving, cancer by smoking, fractured legs by skiing) or events (e.g., volcanic eruptions, earthquakes, explosions).
Other than the measurement of damage, there does not exist a separate method to validate the probability of occurrence (Tittes, 1986; Hauptmanns et al„ 1987; Kaplan and Garrik, 1993). The term probability of occurrence is used for such events of damage where information or even only presumptions about the relative frequency of the event have been given, but where the precise time remains uncertain. Risk statements always describe probabilities, i.e., tendencies of event sequences, which will be expected under specific conditions. The fact that an event is expected on average once each thousand years does not say anything about the time when the event will actually occur.
If we have indications of the determination of the probability of occurrence as well as the extent of damage, we call the degree of reliability of the determination certainty of assessment. If the certainty of assessment is low, one needs to characterize the nature of the uncertainty in terms of statistical confidence intervals, remaining uncertainties (identifiable, but not calculable), and plain ignorance. We use the term uncertainty if we mean the general inability to make deterministic predictions of events of damage (cf. Bon6, 1996). Uncertainty is a fundamental characteristic of risk, whereas the certainty of assessment varies between extremely high and extremely low. Even if it is not possible to make objective predictions about single events of damage on the basis of risk assessment, the assessment is not at all just as you like (Rosa, 1997). When we have two options of action where the same unwelcome event will occur with different probability, the conclusion for a decision under uncertainty is clear: Each rationally thinking human being would choose the option of action with the lower probability of occurrence (Renn, 1996).
From this point of view we consider it to be justified and necessary that technical and natural scientific assessments and social risk perceptions be brought together within rational risk evaluations (Fiorino, 1989). Now the question arises of how societies should decide on fundamental procedures concerning uncertain consequences of collective risks. Which strategy should a society choose if the consequences of risky actions concern many people with different preferences? Philosophers and decision-making the orists come to very different conclusions (cf. Shrader-Frechette, 1991; Leist and Schaber, 1995; Jonas, 1979; 1990; Rawls, 1971; 1974). We want to emphasize that scientifically evaluated risks and theoretical decision-making assessments have an action-determining function despite remaining uncertainty and ambivalence that cannot be replaced by either intuition or actual acceptance, political feeling, or assessments of interests. This is why we apply scientifically ensured evaluations of the respective risks by choosing appropriate tools of regulation.
Therefore, we distinguish three categories of risks for a practicable and rational risk evaluation (see Fig. 1): the normal area, the intermediate area, and the intolerable area (area of permission) (cf. also Piechowski, 1994). The normal area is characterized by relatively low statistical uncertainty, rather low probability of occurrence, rather low extent of damage, high certainty of assessment, low persistency and ubiquity of risk consequences, and low irreversibility of risk consequences, and the risks also have low complexity or empirically proven adequacy. In this case the objective risk dimensions almost correspond to the scientific risk evaluation. For risks in the normal area we follow the recommendations of decision-making analysts who take a neutral risk attitude as a starting point for collective binding decisions.
The "intermediate area" and the "intolerable area" are more problematic because the risks that go beyond ordinary dimensions. Within these areas the certainty of assessment is low, the statistical uncertainty is high, the potential damage can reach alarming dimensions, and systematic knowledge of consequences is missing. The risks can also generate global, irreversible damages that accumulate for a long time or mobilize or frighten the population in a special manner. A clear statement concerning the validity of the scientific risk evaluation is hardly possible. Risk aversien behavior of is absolutely appropriate because the limits of human ability of knowledge are reached. That is why a weighing risk decision is not any more a priority but a limitation of possibilities of wide-ranging negative surprises. Precautionarily oriented strategies of risk con
Intolerable area n
Intermediate area i
□ Normal I | Intermediate I I Intolerable I | Beyond area I-1 area '-1 area I-1 definition
FIGURE 1 Risk areas. Source: WBGU, German Scientific Advisory Council on Global Change (1999).
trol, models of liability of endangering, general norms of caution, and general aspects of risk avoidance have priority.
We consider it useful to include further criteria of evaluation in the characterization of risks (Kates and Kasperson, 1983; California Environmental Protection Agency, 1994). These criteria can be derived from research studies of risk perception or the way they are used or proposed as assessing criteria in several countries such as Denmark, the Netherlands, and Switzerland (cf. Petringa, 1997; Lofstedt, 1997; Hattis and Minkowitz, 1997; Beroggi et aL 1997; Ilauptmanns, 1997; Poumadere and Mays, 1997; Piechowski, 1994). The following criteria are relevant:
• Ubiquity defines the geographical dispersion of potential damages (intragenerational justice).
• Persistency defines the temporal extension of potential damages (intergenerational justice).
• Irreversibility describes the impossible restoration of the situation to the state before the damage occurred (possible restorations are, e.g., reforestation and cleaning of water).
• Delay effect characterizes a long time of latency between the initial event and the actual impact of damage. The time of latency could be of a physical, chemical, or biological nature.
• Potential of mobilization is understood as violation of individual, social, or cultural interests and values by affected people, generating social conflicts and psychological reactions.
In the relevant studies of risk perception most people associate risks with questions of control, voluntariness, addiction to risk sources, and just allocation of risk and benefit (Jungermann and Slovic, 1993). The assessment of control is covered by the criteria of ubiquity and persistency concerning the physical dimensions, and by the criterion of mobilization concerning the social dimensions. From a collective view voluntariness can hardly be taken into consideration as an assessing criterion for societal risks because our relevant risks will be transferred to others. The addiction to risk sources as a single criterion is nor-matively not useful because it is possible that people get used to unacceptable risks (e.g., accidents by driving). Criteria for distributive justice are more difficult to cover because intersubjective valid standards for measuring justice or injustice are lacking. Less problematic is the question of identity between beneficiaries of activities and people affected by risk. If there is identity, individual risk regulation is useful. In other cases collectivc mechanisms of regulation must be implemented. These can reach from commitments of liability to participation of affected people in decisions or procedures of permission. In most cases a casc-to-case consideration is necessary to clearly find out violation of the thesis of justice.
In summary, our criteria and their ranges arc:
Confidence interval for p: high to low certainty of assessment by assessing the probability of occurrence; Confidence interval for d: high to low certainty of assessment by assessing the extent of damage.
• Ubiquity: local to global dispersion.
• Persistency: short to long removal period.
• Irreversibility: damage cannot be restored to damage can be restored.
• Delay effect: a low to high latency between the initial event and the impact of the damage.
• Potential of mobilization: political relevance to high political relevance.
Theoretically, a huge number of risk types can be identified by these eight criteria. Such a huge number of cases would not be useful for the purpose of developing a comprehensive risk classification. In reality, some criteria are tightly coupled together and other combinations are certainly theoretically possible, but there are no or only a few empirical examples. Answering the question of risk priority, risks with several extreme qualities play a special role. We have chosen a classification where single risks are classified as risk types in which they particularly reach or exceed one of the possible extreme qualities. This classification is derived from Greek mythology.
Events of damages that have a probability of almost one are not relevant for us. High potentials of damages with a probability of nearly one can hardly be assessed as acceptable. Such risks occur seldom. In the same way, a probability that goes toward zero is harmless as long as the associated potential damage is not relevant. It is just a characteristic of risk that the range of damage negatively correlates with the level of probability. The higher the damage the lower the probability.
2.4.1 Risk Type "Sword of Damocles"
According to Greek mythology, Damocles was invited to a banquet by his king. At the table he had to sit under a sharp sword hanging on a wafer-thin thread. Chance and risk are tightly linked up for Damocles and the Sword of Damocles became a symbol for a threatening danger in luck. The myth does not tell about a snapping of the thread with its fatal consequences. The threat rather comes from the possibility that a fatal event could occur for Damocles every time even if the probability is low. Accordingly, this risk type relates to risk sources that have very high potentials for damage and at the same time very low probabilités of occurrence. Many technological risks such as unclear energy, chemical facilities and dams belong to this category.
2.4.2 Risk Type "Cyclops"
The ancient Greeks knew of enormously strong giants who were punished despite their strength by only having a single eye. They were called Cyclopes. With only one eye only one side of reality and no dimensional perspective can be perceived. Concerning risks it is only possible for them to ascertain either the probability of occurrence or the extent of damage while the other side remains uncertain. In the risk type Cyclops the probability of occurrence is largely uncertain whereas the maximum damage can be determined. Some natural events such as floods, earthquakes, volcanic eruptions, and El Niño, but also the appearance of AIDS, belong to this category as long as no or only contradictory information exists.
2.4.3 Risk Type "Pythia"
The Greeks of the antiquity asked their oracles in cases of uncertainty. The most known is the oracle of Delphi with the blind prophetess Pythia. Pythias prophecies were, however, ambiguous. It certainly became clear that a great danger could threaten, but the probability of occurrence, the extent of damage, and the allocation and the form of the damage remained uncertain. Human interventions in ecosystems, technical innovations in biotechnology, and the greenhouse effect belong to this risk type where the extent of changes is still not predictable.
2.4.4 Risk Type "Pandora's Box"
The old Greeks explained many evils and complaints with the myth of Pandora's Box—a box that was brought down to Earth by the beautiful Pandora created by the god Zeus. It contained many evils and complaints. As long as the evils and complaints stayed in the box, no damage at all had to be feared. However, when the box was opened, all evils and complaints were released, which then irreversibly, persistently, and ubiquitously struck the earth. This risk type is characterized by uncertainty in both the probability of occurrence and the extent of damage (only presumptions) and by high persistency. Here, ozone-destroying substances can be quoted as examples.
2.4.5 Risk Type "Cassandra"
Cassandra was a prophetess of Troy who correctly predicted the victory of the Greeks, but whose compatriots did not take her seriously. The risk type Cassandra describes a paradox: the probability of occurrence and the extent of damage are known but produce little immediate concern because the damages will occur after a long time. Of course, risks of the type Cassandra are only interesting if the potential damage and the probability of occurrence are relatively high. This is why this type lies in the intolerable area (area of permission). A high degree of the delay effect is typical, i.e., a long period between the initial event and the impact of the damage. An example of this effect is anthropogenic climate change.
2.4.6 Risk Type "Medusa"
Ancient mythology tells that Medusa was one of three snake-haired sisters, the Gorgons, whose appearance turned the beholder to stone. Like the Gorgon who spread fear and horror as an imaginary mythical figure, some new phenomena have this effect on modern people. Some innovations are rejected although they are
□ Normal I ¡Intermedíate I I Intolerable I I Beyond -area I-' area '-1 area I-1 definition -
Classes Pandora risk class:
'v- of risk Only assumptions are possible as to probability of occurence P and extent of damage £
FIGURE 2 Risk types. Source: WBGU, German Scientific Advisory Council on Global Change (1999).
hardly assessed scientifically as threats. Such phenomenona have a high potential for public mobilization. Medusa was the only sister who was mortal-if we transfer the picture to risk policy-Medusa can be combated by rational arguments, further research, and clarification in public. According to the best knowledge of risk experts, risks of this type fall in the normal area. Because of their specific characteristics, these-risk sources frighten people and induce strong refusal of acceptance. Often a large number of people are affected by these risks but harmful consequences cannot be statistically proved. A typical example is electromagnetic fields.
The main objective of risk classification is to gain an effective and feasible policy tool for the evaluation and the management of risks. The characterization provides a platform for designing specific political strategies and measures for each risk type. The strategies pursue the goal of transforming unacceptable into acceptable risks; i.e., the risks should not be reduced to zero, but they should be reduced to a level such that routine risk management becomes sufficient to ensure safety and integrity. All strategies and respective measures are arranged according to priorities. In the normal case more than one strategy and more than one measure are naturally appropriate and necessary. If resources are limited, strategies and measures should be taken in line with the priority list. The following part lists the prior strategies and the prior measures recommended for each risk type.
Was this article helpful?