For more than a decade some of us have been involved in discussions and arguments about the ethics of risk management. Part of that debate has been about the methods, and applications of prescriptive analysis in making decisions involving public risk. By ‘’prescriptive analysis", I mean any systematic approach - it need not be quantitative-to assist risk managers in making decisions consistent with their value judgements and their information about a problem. No one could seriously object to prescriptive analysis per se, but of course, analysts must use some particular method, and different methods can raise different objections.

In the early days, the ethical debate focused on cost-benefit methods for assessing risks and comparing them to other values. Cost -benefit analysis is subject to a number of serious objections, which will be very briefly described in this paper.

Revealed preference theory tells us to look at real economic data, but if these behaviourist strictures seem excessively rigid, we can modify the evidential requirements to include contingent valuation methods or expressed preference theory.

The cost -benefit conception of value is also individualistic. All values are monetized as individual willingness to pay. The cost-benefit decision rule says that after all risks, values, and preferences for each individual have been assigned monetary values, they should be added together and the alternative chosen that maximises net benefits (or an alternative chosen if benefits outweigh costs).

Cost-benefit analysis has been criticized both for its conception of value, and for its decision rule. In turn, the conception of value has been criticized, both for its monetary standard and for its methodological individualism. Thus, we have three kinds of criticisms to consider:

1.Those directed at the behavioural and monetary standards of value embodied in revealed preference theory.

2.Those directed at the methodological individualisms.

3.Those directed at the decision rule.

Objections to the behavioral and monetary standard suggest, that cost-benefit methods cannot account properly for special or sacred values, or for moral conflict.

The most publicised criticism of cost-benefit analysis is that it puts a dollar value on human life and health or on protecting the environment. It invites us to regard these things as commodities and to show a willingness to trade them for other "consumer"goods, a willingness that many would take, in itself, as a sign of moral corruption. Of course, the defenders of cost-benefit methods argue that deciding what to pay is what risk management is all about; therefore, the only issue is whether we are going to be honest and explicit about it.

The debate has not in my opinion been resolved, but this objection suggests another and deeper criticism of the behavioral and monetary metric of value,that it will surely lead to moral distortions. Some of our more cherished values may be tarnished or cheapened by our showing a willingness to pay for them.

Take sex, for example; it is an important value for which many people sacrifice greatly,but it is not uncommon for these same people to be unwilling to pay for it. Our consumer behaviour does not always reveal our most sincere and reflective thoughts about what is most important; nor can it show the difference between being indifferent, because the alternatives seem to us pretty much equal in value, and being conflicted or unable to choose, perhaps because we find the alternatives incomparable on a single scale of value.

We also live in a world of other actors and choosers, whose behaviour affects our outcomes. Our rational choices, both in the economic and in the political sphere, will often be strategic. I act in part on the basis of my beliefs, about how my behaviour will influence the beliefs and behaviour of others, about how my behaviour will influence another`s beliefs about my beliefs about his/her behaviour, and so on. We simply cannot infer preferences or values directly from observing a rational person`s choice behaviour.

Criticisms of the methodological individualism of cost-benefit analysis call attention to concerns about distributive justice. Simply aggregating risks, costs, and benefits will lead to decisions that will especially disadvantage the poor, rural populations, and( because of the discount rate applied to all monetized values) future generations.

Of course, adjustments can be made to the preference theory, to counter these implications. Unless they are made simply to correct for "income effects", they are either ad hoc modifications, or else they lead in the direction of another method of prescriptive analysis - decision analysis, which I shall discuss presently.

Consider an example that illustrates both the objection about distributive justice and the problem with the cost-benefit decision rule. Suppose we have a given amount of resources and must choose to spend the entire amount on one of two alternative projects. One project would reduce, by a given amount, the average individual risk of premature death and would save, let us say, ten lives over a given period of time.

The second project would reduce, by a greater amount, the average individual risk of premature death in a sparsely populated rural area but would save only five lives over the same period. Assuming all other factors are equal, the cost-benefit decision rule would tell us simply to reduce the risk of the urban population.

Now, I do not mean to suggest that this decision is clearly the morally incorrect one or that our intuitions about distributive justice dictate that we must do otherwise. Indeed, in cases like these, I think it is very difficult to know what to do.

The objection to the cost-benefit decision rule is that it short-circuits the dilemma. Government agencies, that have suggested using a cost-benefit decision rule, have been criticized or sued for abandoning their legislative or political mandate. And for approaching the special values they were created to protect with the cold and detached eye of an accountant looking only for the biggest "bang for the buck." Distributive justice does not carry much weight in analyses.

This criticism may also point to the special nature or sacredness of certain values, which may call for certain symbolic measures. In any case, it suggests the importance of procedural values to which cost-benefit methods are entirely blind,

In the example of the choice between urban and rural risk reduction, perhaps the concerns about distributive justice are best met by adopting a procedure for making the choice that represents the concerns of both sides fairly. Ensuring that decision rules reflect different procedural values is surely of primary importance in resolving many risk disputes.

Thus there are three kinds of ethical objections to cost-benefit methods. The behavioural and monetary conception of value introduces moral distortions, and fails to reflect special and sacred values. The methodological individualism leads to decisions that ignore our concerns for distributive justice, and the decision rule is incompatible with procedural values.

In the discussion of the ethics of risk management, however, cost-benefit analysis is a thing of the past.

Recently, the discussion has shifted to decision analysis, in large part because decision analysis can avoid the criticisms just described. Decision analysis makes a minimum commitment to any conception of value. It is thus flexible in determining how alternatives are weighed and compared.

The decision rule is not an algorithm but a prescription. We can reject it and decision analysis can even prescribe against its own use, in a given situation.

Decision analysis is supposed to be a formalization of common sense, to be applied especially to problems that are too complex to rely on common sense alone to analyze adequately. It is based on the axioms of utility theory, which can be interpreted merely as defining a coherent and consistent set of preferences. We can fill in the values however we determine we should, and we can reflect uncertainties to any degree of accuracy.

I do not mean to suggest that decision analysis as a method, is completely immune to ethical doubts. I do think the criticisms that can be raised are rather recherche and abstract.

Moral problems can certainly arise in any application of decision analysis, as its proponents have been quick to recognise and admit. I think we must agree with Ron Howard (1) that as a method, decision analysis is basically "amoral, like arithmetic", or with Ralph Keeney (2), that decision analysis can be made "consistent with any of the major ethic theories." To see that this is true, consider how the proponent of decision analysis would respond to the three kind of objections I raised against cost-benefit analysis.

Because decision analysis is silent about how we value a state of affairs, we are free to include special values, sacred values, non-marketable values, and so on. The only thing we are required to do is compare, and rank different alternatives consistently. That can be done without prejudice to the nature of our concerns.

Even if we take some values as absolute, unable to be traded off without being compromised, decision analysis can reflect such rankings too. Decision analysis may show us how difficult it is in practice to regard some values in this way and to remain constant; however the analysis can do this with a certain "lightness of step" without being condescending, nagging, or simply obtuse.

The minimal substantive moral commitment of decision analysis also means that it is not committed, as is cost-benefit analysis, to methodological individualism. If a state of affairs includes non-reducibly social values, these can be included.

If we value certain moral rights, for whatever reason, then they contribute value to the state of affairs in which they are recognised and protected.

If we value fair distributions, then we can likewise include distributions of risks, costs, and benefits among the attributes that add value to a state of affairs. We can do this either through weighing the outcomes to some individuals or groups more heavily than to others, and then aggregrating them, or we can do this by measuring the fairness or unfairness of the distribution of risks or consequences in a state of affairs directly and including fair distribution itself as an attribute of utility.

Finally, decision analysis allows us to count a state of affairs better in which a decision is reached by using a favoured procedure rather than a state of affairs in which otherwise identical consequences are realized in a different manner.

Thus, if we must choose between saving the lives of ten urbanites or the lives of five ruralites, we could decide that the best outcome of all is to invoke a fair procedure that determines that we shall save the ten lives. We can decide that the outcome of invoking the procedure and saving five lives is ethically better than saving ten lives without invoking the procedure. Which outcome we value more will depend ultimately on what we think about the importance of invoking the procedures as compared with the importance of risking the loss of the five extra lives, but decision analysis will allow us to go either way on this choice.

The more one comes to appreciate the flexibility of decision analysis, the more convincing is the argument that it is indeed substantially ethically neutral. However, another conclusion to draw from these arguments, as the example about procedural values demonstrates clearly, is that decision analysis does not help us at all in resolving ethical conflicts or making moral progress in risk management because it gives us no moral guidance. The only value it promotes is consistency, respect for which may well constitute progress over our normal practices but hardly scratches the surface of our ethical concerns.

Once we become convinced that methods of prescriptive analysis need not be ethically suspect, the moral debate has been neither concluded nor even advanced. Everything remains to be done, and what remains is a lot.

Let us look one final time at the three kinds of objections to cost-benefit analysis. The first you will recall, was the behaviourist and monetary conception of value, which turns out to be morally distorting because it is insensitive to a range of special and important values. Decision analysis allows us to be sensitive to these values, but it does not tell us how to be sensitive to them, and that is our problem.

What are we to make of values that seem compromised when we are asked to determine their economic worth? We know that people feel this way about a number of the things that are important subjects for risk management. Our surveys on risk attitudes tell us this, and the few anthropological studies in this area may help to explain them and their role in the fabric we call culture. However, we still do not know how risk managers should tiptoe around these issues, or what role a government agency may be expected to fill in giving these values the symbolic or expressive actions they demand; and lives, as well as serious economic consequences are at stake.

Similarly, even if one method of prescriptive analysis fails to reflect the concerns of distributive justice and another one does not essentially fail in this regard, we still need to know more about the values of justice relevant to risk management. In particular, there are two issues we need to better understand. The first is the connection between fairness and procedures.

If we invoke fair procedures, or even if we distribute risks equally, to what extent does this relieve us from having to be concerned with unequal outcomes? When are those outcomes unfair or unjust?

This question points to the role of compensation, which suggests a second central question about justice. Part of understanding the special nature of certain values or the qualitative differences among values is understanding the limits of the ability compensation to achieve justice.

Compensation can fail in two different ways for different reasons, either because a benefit fails to equal in amount the loss for which it is meant to compensate or because it fails to offset the loss because it is a failure in the kind of benefit required.

When we say that a loss cannot be compensated, we can mean either that it is too great to be offset or that it cannot be offset by the nature of the value. The importance of understanding this difference will be to understand when risks can be imposed, if the losses are compensated, and when risk-imposing activities should be prohibited morally.

Finally, we again see similar issues arising when we look for morally adequate decision-making procedures. In risk-management, these will often involve a requirement to involve affected parties in the decision-making process, to secure their free and informed consent.

Again, the kind of risk research we have from the social sciences does more to underscore the problems than to suggest solutions. Here we have to confront the limitations in human judgement, our tendency towards preference reversals (which raises deep problems about the coherence of our values), and the difficulties of finding effective and non-biasing ways to communicate technical and expert information.

What are we to make of all these problems? I do not know, and I do not think anyone else knows very well either. I would like to see some progress made on the ethical issues in risk-management, and perhaps we shall when we can put aside our doubts about prescriptive analysis per se. I would like to contribute to this progress in our moral understanding; however, we all need to provide more support than this subject has received.