How safe is safe enough?

Abstract

It is more or less axiomatic in the Journal of Flood Risk Management that flood management decisions should be based upon risk. Flood risk management decisions involve the identification of a range of possible flood risk management measures (structural and non-structural) and evaluating their potential benefits, in terms of risk reduction, now and in the future. Then comes the difficult decision: by how much should risk be reduced? Broadly there are two potential approaches to addressing this problem: cost-effectiveness analysis and cost–benefit analysis. Both are fraught with difficulties. Cost-effectiveness analysis seeks to identify the least-cost option that satisfies some performance requirement. In flood risk management, the primary performance requirement will be in terms of acceptable risk, although there will often be other constraints in terms of environmental standards and socio-political acceptability of the proposed measures. Definition of acceptable risk is a thorny issue. Different countries have very different approaches, and indeed cultural reactions, to the very notion of acceptable risk. In the Netherlands analysis of societal risks in terms of F-N curves is well established. Standards of dike reliability are established for all dike rings, ranging from 1:10 000 years on the coast to 1:1250 years on the Rhine river channels. The United Kingdom explored ideas of acceptable societal risk, notably during the notoriously long-winded public enquiry for the Sizewell B nuclear power station but then retreated from fixed thresholds for acceptable risk to the more flexible ‘As Low As Reasonably Practicable’ (ARARP) framework. In practice, that means that there is a range of risk thresholds derived from government guidance, insurance availability and implicit social contract. Guidance for development in flood plains introduces thresholds at 1:100 years for river flooding and 1:200 years for coastal flooding. Meanwhile, availability of domestic flood insurance provides another threshold of public acceptability. In the United Kingdom the now fragile ‘statement of principles on flood insurance’, which sets out the conditions in which insurance companies will, for the time being at least, offer flood insurance domestic customers, implies an acceptable probability of flooding of about 1:75 years. A flood probability greater than this will not be acceptable because flood insurance would be unavailable to new customers, so they would not be able to obtain a mortgage on their home. A similar relationship between flood probability and insurance availability exists in the United States where levees have to be certified for Federal flood insurance to be made available. A further criterion for acceptability relates to business confidence. In 2008 the Dutch Deltacommissie proposed an increase in standards of flood protection by a factor of 10. Although flood defence standards were already the highest in the world, in the aftermath of Hurricane Katrina and amidst new scientific evidence about the possibilities of sea level rise, this proposed increase in standard was a clear signal to the world that the Netherlands would be in business come what may. It is not hard to see how an emphasis on absolute thresholds may lead to perverse decisions. The cost of reducing the probability of flooding from just over the threshold to just under may be excessive; is it worth it? An acceptable probability threshold does not necessarily account for the severity of the consequences of flooding, which, from a risk perspective, we should be taking account of. Cost–benefit analysis provides a more rational framework for the setting of risk standards. Society should, and from time to time does, object to disproportionate sums being spent on achieving high standards of safety when those resources could be spent more effectively at reducing another risk or indeed providing other benefits to society. Cost–benefit analysis provides a rational framework for making these resource allocation choices and has become the dominant approach to the appraisal of public investment schemes in many countries worldwide. Yet cost–benefit assessment is fraught with well-known methodological problems, in particular in relation to the proper valuation of loss of life and social and environmental impacts. Increasingly we aim to deliver flood risk management schemes that yield other benefits, for example with respect to waterfront development or habitat restoration, but valuation of those benefits is challenging. Discounting means that costs and benefits that are incurred a long time in the future can have a negligible effect on cost–benefit decisions, even though we can foresee situations in which future generations might regret decisions taken by present generations, for example in relation to allowing development in vulnerable locations. Intergenerational issues are one version of the many equity issues that cost–benefit analysis raises; who incurs risks and benefits, and who pays? It is particularly tricky to estimate the opportunity costs of foregone development in flood plains. In this Journal we would argue that prudent flood-plain zoning policies are an essential aspect of flood risk management. But we must not overlook the fact that these policies limit development opportunities and so brings a cost to society. Yet it is almost impossibly difficult to evaluate the wider societal benefits of development that would have materialised had development been permitted. In theory cost–benefit analysis would allow the identification of an ‘optimal’ level of risk, now and in the future. In practice the multiple perspectives on risks mean that it is arguable as to whether a unique optimum exists. The attendant uncertainties mean that it would be impossible to identify it precisely even if it did exist. What cost–benefit analysis can provide is evidence to inform resource allocation decisions and a mechanism to avoid the extremes of under- or over-adaptation. Cost–benefit analysis is open to distortion (for example when it is being used to justify the expenditure of other people's money), but at least it provides a quantified framework that can be opened up to scrutiny in which assumptions, data and calculations are challenged. We need to be realistic about what acceptable risk thresholds and cost–benefit analysis can and cannot provide. Our emphasis is upon improving the process as well as the material outcomes of flood risk management decisions. An important principle in flood risk management is that the depth of analysis should be in proportion to the severity of the risk and the magnitude of the mitigation decision. Not all flood risks will require a detailed analysis, but on the whole we could do much better in evaluating risks, costs, and attendant uncertainties and in communicating them to decision makers and other stakeholders. The Journal of Flood Risk Management seeks to contribute to this endeavour. Jim Hall Associate Editor