Expert judgement elicitation for risk assessments of critical infrastructures

Governmental bodies and companies are confronted with the problem of achieving rational consensus in the face of substantial uncertainties. The subject area of this special issue (risk and vulnerability assessments and management of critical infrastructures) might be a good example as are risk management of chemical installations and accident consequence management for nuclear power plants. Decisions with regard to infrastructures functioning and possible malfunctioning must be taken on the basis of predictions of technical and organizational system behaviour. These predictions use mathematical models containing scores of uncertain parameters. Decision makers want to take, and want to be perceived to take, these decisions in a rational manner. The question is, how can this be accomplished in the face of large uncertainties? One available source is experts in the many fields of interest within infrastructures. This paper describes the use of structured expert judgement in a formal manner. The paper refers to the Procedures Guide published by the European Union as EUR 18820. This Procedures Guide addresses two methods for using expert judgements developed at Delft University of Technology. The paired comparisons method is particularly suitable to identify the relative importance of attributes in the risk management arena, while the Classical Model, apt to arrive at subjective probability assessments, is particularly suitable to derive uncertainty distributions over model parameters. Examples will be referred to for further illustration of applications relevant in the field of risk assessment and risk management.     

Expert judgements of sea-level rise at the local scale

Whilst local projections of sea-level rise (SLR) are necessary to facilitate targeted climate change adaptation and communication strategies, downscaling from global climate models can be problematic. Here, we use expert probability judgement to elicit a suite of local projections, and associated uncertainties, for future SLR on the Severn Estuary in the south-west of the UK. Eleven experts from a range of policy and academic backgrounds took part in a structured probability elicitation exercise for the years 2050, 2100 and 2200. In addition to the quantitative elicitation, the experts’ reasoning during the task was qualitatively analysed. Quantitative analyses show that although there is consensus that sea levels will rise on the Estuary in future, there is wide variation between judgements and much uncertainty regarding the magnitude of future rise. For example, median estimates of SLR (compared to the 2011 level) range from 9.6 to 40 cm in the year 2050; 20 to 100 cm in 2100; and 35 to 300 cm in 2200. Fifty per cent confidence intervals and ninety per cent confidence intervals vary even more. Qualitative analyses indicate that experts’ judgements may have been influenced by their choice of methods and information sources, the ways in which they thought about the future, and heuristics. The study shows the merits of integrating qualitative and quantitative methods to explore the reasoning behind uncertainty judgements. We conclude that where expert probability judgements are to be used to characterise uncertainty such reasoning should be made explicit.     

The constitutive element of probabilistic agency in risk: a semantic analysis of risk,danger, chance,and hazard

Defining central concepts with accuracy is crucial to any scientific discipline. A recent debate over risk definitions in this journal illustrates the far reaching consequences of divergent definitions. Aven and Renn define risk as a social construct while Rosa defines risk as an ontological fact. Both claim that their definition reflects the common usage of the word risk. Through a semantic analysis this paper points to a constitutive element of what is termed probabilistic agency in the risk concept. In this respect, risk is distinct from danger, and because Rosa’s main argument is based on the apparent synonymy between risk and danger, the premises for his risk ontology are not valid. The paper furthermore argues that Aven and Renn’s attempt to bridge between epistemology and ontology is based on a distinction between a conceptual level of risk and its practical application which is impossible to uphold if a risk definition is to be in accordance with the ordinary usage of the word. The paper concludes by arguing that risks are only real within a subjective ontology.     

The relationship between liquidity risk and credit risk in banks

This paper investigates the relationship between the two major sources of bank default risk: liquidity risk and credit risk. We use a sample of virtually all US commercial banks during the period 1998–2010 to analyze the relationship between these two risk sources on the bank institutional-level and how this relationship influences banks’ probabilities of default (PD). Our results show that both risk categories do not have an economically meaningful reciprocal contemporaneous or time-lagged relationship. However, they do influence banks’ probability of default. This effect is twofold: whereas both risks separately increase the PD, the influence of their interaction depends on the overall level of bank risk and can either aggravate or mitigate default risk. These results provide new insights into the understanding of bank risk and serve as an underpinning for recent regulatory efforts aimed at strengthening banks (joint) risk management of liquidity and credit risks.     

Expert Judgments Regarding Risks Associated with Salmon Aquaculture Practices in British Columbia

Making sound decisions about managing ecological risks necessarily involves relying on judgments by technical specialists informed by the best available scientific evidence. Yet, organizing those judgments in ways to assess the relative risks of different components of a technology, and considering priorities in managing those risks, is a difficult and under‐explored aspect of environmental management. In this study, we elicited the judgments of scientists associated with the salmon aquaculture industry in British Columbia in order to learn their expert viewpoints of potential risks. This paper presents survey results regarding structured judgments provided by scientists engaged in studies associated with aquaculture or preserving wild stocks of Pacific salmon species. There were statistically significant differences regarding judgments of the risks of various current aquaculture practices on wild salmon stocks. It was possible to rank the means of scientific judgment scores to prioritize these risks. Differences in rankings were location and context specific.     

Determining overall risk

The risk journal literature lacks a clear and simple account of the conceptual issues involved in determining the overall risk of an action, and in explaining how risk is additive. This article attempts to bring a measure of clarity to these issues in as basic and non‐technical a way as possible. First of all, the view that risk is ‘expected harm’ is explained. The view that risk is a quantitative concept is then defended. The distinction between the risk run by doing action A in respect of possible outcome x, and the overall risk run by doing action A in general is explained, as is the position that the overall risk of A is determined by summing the risks of each possible harm that A could give rise to. The article then explains how risks can be summed over time, as long as the probabilities involved are determined according to probability theory. Finally, the article explains that in a doing a risk‐benefit analysis of A, positive aspects of a possible outcome x, where x is harmful on balance, must be incorporated into x's level of harm rather than incorporated into the benefit side of the risk‐benefit analysis of A.     

First-passage probability, jump models, and intra-horizon risk

This paper proposes a risk measure, based on first-passage probability, which reflects intra-horizon risk in jump models with finite or infinite jump activity. Our empirical investigation shows, first, that the proposed risk measure consistently exceeds the benchmark value-at-risk (VaR). Second, jump risk tends to amplify intra-horizon risk. Third, we find large variation in our risk measure across jump models, indicative of model risk. Fourth, among the jump models we consider, the finite-moment log-stable model provides the most conservative risk estimates. Fifth, imposing more stringent VaR levels accentuates the impact of intra-horizon risk in jump models. Finally, using an alternative benchmark VaR does not dilute the role of intra-horizon risk. Overall, we contribute by showing that ignoring intra-horizon risk can lead to underestimation of risk exposures.     

Expert relevance and the use of context-driven heuristic processes in risk perception

The effectiveness of a medical treatment should not predict its risk (highly effective treatments can be either safe or risky), however, people’s use of heuristic shortcuts may lead them to judge a link between effectiveness and risk, typically a negative correlation. A particular concern is that experts might use such a strategy and that this is unlikely to provide an accurate judgement. This large-scale field-based experiment compares expert-relevant and non-expert-relevant contexts, for both expert and public judgements of risk and effectiveness in the context of blood transfusion medicine. Postal questionnaires were distributed to anaesthetists (experts, N?=?123) and a general public (non-expert) comparison group (N?=?1153); half of the participants were cued with accompanying general information about genetically-modified (GM) biotechnology and half received specific information about blood product technologies. The blood-focussed information served to emphasise the medical relevance of the questionnaire to the expert group. Regression analyses showed that generally perceived effectiveness predicted perceived risk for both experts and non-experts, which suggests heuristic processing. However, although experts appeared to engage in heuristic processing for risk perceptions in certain circumstances, this processing is strongly affected by context. Experts who received the medically relevant context rated perceptions of effectiveness independently of perceptions of risk, unlike those who received the GM context. This indicates a reduced reliance on a low-effort heuristic for experts given an expertise-relevant context. The results are considered in light of dual-process (rational-associative) accounts of reasoning.     

The challenges of risk society for impact assessment

This paper takes its point of departure in Ulrich Beck’s theory of risk society and the aspects that characterise this society. The paper puts forward a hypothesis, on which theoretical challenges the characteristics of risk society pose to impact assessment as a decision support tool; namely, the challenge of delivering assessments and predictions and the challenge of handling differences of opinion and debate. Through a case example of integration of climate change in strategic environmental assessment, the paper uses empirical evidence from a survey and a series of interviews to carry out a preliminary discussion of how the theoretical challenges are reflected in practice. The case study results show that the challenge of delivering assessments and predictions in a risk society is reflected in the current state of practice, while the challenge of handling differences of opinion and debate is not clearly reflected.     

Sovereign risk assessment and agency credit ratings

Sovereign ratings are gaining importance as more governments with greater default risk borrow in international bond markets. However, while the ratings have proved useful to governments seeking market access, the difficulty of assessing sovereign risk has led to agency disagreements and public controversy over specific rating assignments. Recognising this difficulty, the financial markets have shown some scepticism toward sovereign ratings when pricing issues.     

Asymptotic ruin probabilities for a discrete-time risk model with dependent insurance and financial risks

This contribution focuses on a discrete-time risk model in which both insurance risk and financial risk are taken into account and they are equipped with a wide type of dependence structure. We derive precise asymptotic formulas for the ruin probabilities when the insurance risk has a dominatedly varying tail. In the special case of regular variation, the corresponding formula is proved to be uniform for the time horizon.     

On long-term credit risk assessment and rating: towards a new set of models

Institutional investors are supposed to assess credit risk by using a combination of quantitative information such as option models and qualitative assessments. Although option models can be easily constructed, they are not so suitable for the assessment of long-term credit risk that is required by institutional investors. This is mainly because the probability of bankruptcy varies so widely depending on the timing of assessment. We propose a new set of assessment models for long-term credit risk which does not necessarily use stock prices and may incorporate business cycles. The new grand model consists of the two pillars: a long-term cash flow prediction model and a credit risk spread assessment model. The calculated values derived from these models are effectively usable for reasonable calculation of risk spreads. It is quite interesting to see that our investigation indicates that rating bias may exist in the credit risk assessment of the market.     

Rhetorical accounts of risk: interprofessional risk assessment in operational planning meetings

In the context of high-risk industries, risk assessment takes place not only through standardized methods for risk analysis, but is frequently negotiated and discussed as an integral part of operational decision-making. This is not least the case in the context of operational planning. Frequent changes in operations require ongoing assessment of risk as tasks are rescheduled and resources reallocated. The current study explores how professionals account for the presence or absence of risk in a setting in which risk analysis is not the primary objective. With data from the offshore petroleum industry, the rhetorical aspects of risk assessments are examined. A series of interprofessional planning meetings were video recorded, transcribed, and analyzed using a rhetorical discourse analytic framework. The data are analyzed at a micro-interactional level in order to study how accounts of risk are presented and negotiated in this particular setting. The meaning and consequences of operational plan changes, and their implications for safety, are seen as negotiated discursively through interprofessional meeting talk. The analysis shows that accounts of risk are characterized by shifting rhetorical strategies that can be heard to echo established risk discourses often referred to as ‘technico-scientific’ and ‘contextualized’ conceptions of risk. Rhetorical devices are used interchangeably and strategically by the participants as they account for risk from their respective institutional positions and their specific areas of expertise and responsibility. The accounts are found to be increasingly persuasive and rhetorical in style as disagreements over risk and prioritizations surface. Accounts of risk, then, are not simply objective presentations of probability and consequence, but rather powerful tools for achieving specific professional outcomes. The study contributes to the understanding of risk assessment at its most concrete and practical level; as it takes place through professional interaction in an operational setting.     

A risk assessment model for banks

Summary. This paper proposes a model to assess risk for banks. Its main innovation is to incorporate endogenous interaction among banks, where the actual risk an individual bank bears also depends on its interaction with other banks and investors. We develop a two-period general equilibrium model with three active heterogeneous banks, incomplete markets, and endogenous default. The model is calibrated against UK banking data and therefore can be implemented as a risk assessment tool for regulators and central banks. We address the impact of monetary and regulatory policy, credit and capital shocks in the real and financial sectors.We are grateful to Lea Zicchino, an anonymous referee, seminar participants at the Bank of England, and the third Conference in Research in Economic Theory and Econometrics, Syros, for helpful comments. The views expressed are those of the authors and do not necessarily reflect those of the Bank of England.This revised version was published online in January 2005 with corrections to the Cover date.     

Credit risk assessment of fixed income portfolios using explicit expressions

We propose a model to assess the credit risk features of fixed income portfolios assuming they can be characterized by two parameters: their default probability and their default correlation. We rely on explicit expressions to assess their credit risk and demonstrate the benefits of our approach in a complex leveraged structure example. We show that using expected loss as a proxy for credit risk is misleading as it does not capture the dispersion effects introduced by correlation. The implications of these findings are relevant for improving current risk management practices and for regulation purposes.     

On semiparametric estimation of ruin probabilities in the classical risk model

The ruin probability of an insurance company is a central topic in risk theory. We consider the classical Poisson risk model when the claim size distribution and the Poisson arrival rate are unknown. Given a sample of inter-arrival times and corresponding claims, we propose a semiparametric estimator of the ruin probability. We establish properties of strong consistency and asymptotic normality of the estimator and study bootstrap confidence bands. Further, we present a simulation example in order to investigate the finite sample properties of the proposed estimator.     

Crisis risk and risk management

We assess the state of knowledge about crisis risk and its implications for risk management. Data that became available after the global financial crisis show that some types of crises are predictable when accounting for interactions between risks. However, other types of crises do not seem predictable. There is no evidence that the frequency of economic and financial crises is increasing. While data show that an economic crisis is more likely following a political crisis, there is no comparable evidence for climate events. Strategies that increase firm operational and financial flexibility reduce the adverse impact of crises on firms.     

The influence of the forecast horizon on judgemental probability forecasts of exchange rate movements

An experiment is reported which compares directional forecasting performance of experts, novices and simple statistical models over three time horizons on a task involving probabilistic forecasts of exchange rate movements. Probability-judgement accuracy analyses illustrated no clear overall performance differences between experts and novices, but significant differences between the groups on various important components of judgement suggested that the groups obtained their similar overall scores using different cognitive strategies. Striking horizon effects and expertize–horizon interactions were also observed. The subjects performed better than a random walk forecaster, but worse than the random walk with constant drift and first-order autoregressive models. Composite human judgement, however, not only improved on individual judgement but, also, surpassed the simple statistical models in many instances. Possible explanations are offered for these results, suggestions are made for future research, and practical implications are emphasized.     

Risk assessment and risk decision-making process related to hazardous installation in France

The chemical process industry gathers critical infrastructures since chemical plants represent for the society both an instrument to generate activity and benefits (production of goods, employment, services …), and an instrument that can harm people and the environment. This paper describes the risk assessment approach need in France for licensing hazardous installations, in particular those covered by the Seveso II Directive and the implication for land-use planning. The first part explains why the French risk assessment procedure is based on the quantitative evaluation of major accident scenarios with explicit criteria, and only on qualitative analysis of the risk reducing measures without explicit criteria. The second part shows some on-going evolutions in the risk assessment process in particular for the analysis of the safety barriers. In the third part, the authors share some thoughts on the risk decision-making process.     

On a risk measure inspired from the ruin probability and the expected deficit at ruin

In this paper, we study a risk measure derived from ruin theory defined as the amount of capital needed to cope in expectation with the first occurrence of a ruin event. Specifically, within the compound Poisson model, we investigate some properties of this risk measure with respect to the stochastic ordering of claim severities. Particular situations where combining risks yield diversification benefits are identified. Closed form expressions and upper bounds are also provided for certain claim severities.