Risk Meter Research Articles

Assessing network risk with FRM: links with pricing kernel volatility and application to cryptocurrencies

The Financial Risk Meter (FRM) employs Quantile-LASSO regression to identify systemic financial risk and dependencies among tail events across financial assets. This paper establishes, both theoretically and empirically, a meaningful economic relationship between the FRM index, derived from the penalization parameter in quantile LASSO regression, and the volatility of assets' pricing kernels, the attainable maximal Sharpe ratio, and market volatility. Despite the rapid growth of the crypto market and its increasing integration with traditional financial markets, there remains a dearth of risk measures in this space. FRM @ Crypto exhibits robust predictive capabilities in anticipating future market risk, potentially filling a critical void in this market.

Assessing the Impact of Ground Motion Duration on Losses in Typical Modern Steel Moment Frames

This research was undertaken to study the duration effects on the seismic economic risk of steel moment frame (SMF) buildings, a prominent class of buildings in commercial stock. Firstly, a modified version of FEMA P-695 ground motion scaling, tailored for seismic loss estimation purposes and incorporating two sets of spectrally matched bi-directional short- and long-duration ground motions, is proposed to study code-compliant plan-symmetrical SMFs with different heights (i.e., two to 20 stories). It is shown that long-duration ground motions increase the collapse risk of SMFs, on average, by 28.0% at the MCE level. Next, a component-based loss estimation methodology was adopted for evaluating the seismic losses under each set of ground motions. These losses are studied separately for building components (i.e., structural and nonstructural) and contents. Moreover, we propose an approach for calculating average annualized loss (AAL) as a prominent risk meter that segregates contributions of short- and long-duration ground motions to attain hazard consistency. Loss analyses showed the minimal impact of building height on the contribution of these two types of earthquakes. The seismic risk analysis of buildings also revealed that collapse risk is influenced mainly by duration effects followed by building and content losses.

ПЕРСОНИФИЦИРОВАННЫЙ ПОДХОД К ОЦЕНКЕ РИСКА РАЗВИТИЯ ПОВТОРНЫХ СЕРДЕЧНО-СОСУДИСТЫХ СОБЫТИЙ И ХРОНИЧЕСКОЙ БОЛЕЗНИ ПОЧЕК У ПАЦИЕНТОВ С ИНФАРКТОМ МИОКАРДА И ОСТРЫМ ПОВРЕЖДЕНИЕМ ПОЧЕК

Introduction: AKI (acute kidney injury) is a common comorbid condition in patients with AMI (acute myocardial infarction), complicating its course and treatment. In the future, the prognosis of these patients after discharge from the hospital, as well as the incidence of recurrent cardiovascular events (CVS) and chronic kidney disease (CKD) are not known. The purpose of the study: to study the predictors of the development of recurrent cardiovascular events (CVS) in patients with myocardial infarction (MI) with acute kidney injury (AKI). Develop a calculator - a risk meter for repeated CCC. Materials and methods: the study was performed in the Altai Regional Cardiological Dispensary. 193 patients with MI and PCI (percutaneous coronary intervention) were included, divided into 2 groups: the first - 123 patients aged 62.8±1.1 years with MI and AKI, the second - 70 patients without AKI, age 61.3± 1.6 years. Upon admission and before discharge, the level of the KIM-1 molecule (Kidney Injury Molecule-1) and IL-18 (interleukin-18) was studied by ELISA in the urine. All patients underwent coronary angiography using low-osmolar contrast. One year after discharge, the frequency of development of repeated CCC was assessed. Statistical calculations were carried out using statistical packages STATISTICA 12.0. The level of statistical significance was taken equal to 0.05. Results. The level of KIM-1 (at admission) in the first group was statistically higher than in the second: 1998.9±147.6 versus 1289.8±126.1 pg/ml p=0.001, and IL-18 in the AKI group exceeded the corresponding parameter of the group without AKI: 179.0±12.9 versus 114.9±11.5 pg/ml p=0.007, respectively. A year later, it was found that the frequency of recurrent cardiovascular events was higher in the group with AMI and AKI: unstable angina was diagnosed in 26 (21.1%) patients in the group with AKI and 5 (7.1%) patients without AKI in history, p=0.010; 19 (15.4%) and 2 (2.8%) patients, p=0.006, 20 (16.2%) and 3 (4.2%), p=0.013 patients, respectively, had repeated myocardial infarction and progression of CHF in a year . Progression of renal dysfunction during the year was observed 5 times more often in patients with MI and AKI: CKD C2 was diagnosed in 49 (39.8%) patients of the first group and 16 (22.8%) of the second, p = 0.01; CKD C3a and C3b stages - in 34 (27.6%), p˂0.001 and 19 (15.4%) patients, p=0.002, respectively, CKD stage C4 was detected in 4 (3.2%) patients of the group with AMI and AKI in the early postinfarction period. As a result of constructing a multifactorial logistic regression model, predictors were identified that have a multiplicative effect on the development of recurrent CV events: BMI over 25 kg/m2 increases the risk of recurrent CV events by 0.91 [0.83; 0.99] times, p=0.028, history of prior MI in 3.32 [1.24; 9.86] times, p=0.022. Increase in CRP by 1.01 [1; 1.03] times, p=0.045, troponin I, 0.97 [0.94;1] times, p=0.037, and IMT-1, 1 [1;1] times, p=0.030. On the development of CKD: an increase in age is associated with an increase in the chances of developing CKD by 1.18 [1.1; 1.29], p ˂0.001; troponin I by 0.96 times [0.92; 1], p=0.030; systolic blood pressure in 1.03 [1; 1.07] times, p = 0.029. A calculator has been developed for risk stratification of the development of recurrent CV events and an algorithm for managing this category of patients after discharge from the hospital. Conclusions: thus, the identified predictors make it possible to calculate the risk of developing recurrent CV events and CKD using the developed calculator and allocate these patients to a separate group for personalized observation, taking appropriate measures of secondary prevention.

A Systematic Literature Review on Cyber Threat Intelligence for Organizational Cybersecurity Resilience.

Cybersecurity is a significant concern for businesses worldwide, as cybercriminals target business data and system resources. Cyber threat intelligence (CTI) enhances organizational cybersecurity resilience by obtaining, processing, evaluating, and disseminating information about potential risks and opportunities inside the cyber domain. This research investigates how companies can employ CTI to improve their precautionary measures against security breaches. The study follows a systematic review methodology, including selecting primary studies based on specific criteria and quality valuation of the selected papers. As a result, a comprehensive framework is proposed for implementing CTI in organizations. The proposed framework is comprised of a knowledge base, detection models, and visualization dashboards. The detection model layer consists of behavior-based, signature-based, and anomaly-based detection. In contrast, the knowledge base layer contains information resources on possible threats, vulnerabilities, and dangers to key assets. The visualization dashboard layer provides an overview of key metrics related to cyber threats, such as an organizational risk meter, the number of attacks detected, types of attacks, and their severity level. This relevant systematic study also provides insight for future studies, such as how organizations can tailor their approach to their needs and resources to facilitate more effective collaboration between stakeholders while navigating legal/regulatory constraints related to information sharing.

A financial risk meter for China

The study presents FRM@China, a novel risk meter that uses quantile-lasso regression to detect systemic financial risk and dependencies of tail-events among critical financial institutions in China. The analysis demonstrates the high level of robustness of FRM@China in predicting tail-event risks and a negative correlation between FRM@China and FIs' TE risk interconnectedness. To address limitations of the current FRM approach, the study employs the Shapley value to assess the impact of different macroeconomic variables. The results offer policymakers a valuable tool for monitoring market liquidity and evaluating financial policy responses to prevent severe financial risks.

Systemic risk and financial system network using financial risk meter: the case of Vietnam

ABSTRACT In the financial market, systemic risk is defined as the possibility that an event at the company level could trigger severe instability or collapse of an entire industry or the whole economy. Thus, understanding systemic risk is crucial for the financial institutions, large corporations, investors and regulators. This article investigates systemic risk and spillover effect using the new Financial Risk Meter ( F R M ) index, which is obtained from running quantile linear regression and Least Absolute Shrinkage and Selection Operator ( L A S S O ) method. The F R M index is obtained to identify the highly risky periods, the contributors to systemic risk and the potential activators of spillover effect. Moreover, interconnection between firms can be visualized as a network. We use a data set consisting of daily stock returns from 35 financial institutions and real estate firms in Vietnam, combined with 4 macroeconomic variables over the period from November 2011 to December 2020. The findings indicate that over the considered period, some detected highly risky periods are 2012, 2018 and 2020, probably due to the non-performing loan crisis in Vietnam, US-China trade war and global COVID-19 outbreak. Some active activators of risk spillover effect are also identified.

Quantitative Risk Assessment and Management of National Defense Acquisition with a Game-Theoretic Security Risk Meter Tool

Quantitative Risk Assessment and Management of National Defense Acquisition with a Game-Theoretic Security Risk Meter Tool

FINANCIAL RISK METER FOR CRYPTOCURRENCIES AND TAIL RISK NETWORK-BASED PORTFOLIO CONSTRUCTION

Cryptocurrencies have emerged as a new asset class. In order to provide a thorough understanding of this new asset class, we study the dependencies in tail risk events within cryptocurrencies, and provide a hedging alternative in this paper. First, we adopt the Financial Risk Meter approach for cryptocurrencies, which is able to identify individual risk characteristics and indicate systemic risk in a network topology. Next, we detect the interdependencies across digital coins and study the spillover effects. Finally, we construct tail event sensitive portfolios and test the performance versus traditional approaches from January 2019 to May 2022.

Identification of the risk of IHD development in young people, formation of a diagnostic model

Background and Aims : To study the level of biomolecules characterizing the main pathogenetic links of coronary atherosclerosis. To develop a model for diagnosing the risk of coronary heart disease in young people based on significant biochemical parameters.Methods: The study included 49 people with IHD and 127 people without IHD. All participants signed an Informed Consent. A survey was conducted using validated questionnaires. Anamnestic data, anthropometry, functional examinations were collected from the all participants. Total antioxidant activity, MDA-oxLDL, SOD, MCP-1, TNF-alpha, IL-1, IL-6, IL-8, total cholesterol, TG, HDL and LDL were determined in blood serum using enzyme immunoassay, multiplex assay and enzymatic method.Results: The analysis of the data obtained revealed a difference in concentrations between the studied subgroups for the total antioxidant activity, SOD, MCP-1 and IL-6. In patients with IHD, the level of the total antioxidant activity was significantly lower by 2.9 times and MCP-1 by 1.4 times, the concentration of SOD and IL-6 was higher by 16 and 30%, respectively. For interpretation, we used a scatter diagram of the density of values - SHAP, reflecting the degree of influence of each parameter on the resulting variable. The greatest contributions to the model project were made by: the total antioxidant activity, SOD, MCP-1, IL-6 and HDL.Conclusions: The data obtained by us can be recommended for the development of a coronary heart disease risk meter in young people, which includes the most significant laboratory parameters. This study was conducted within the framework of the project RSF 21-15-00022. Background and Aims : To study the level of biomolecules characterizing the main pathogenetic links of coronary atherosclerosis. To develop a model for diagnosing the risk of coronary heart disease in young people based on significant biochemical parameters. Methods: The study included 49 people with IHD and 127 people without IHD. All participants signed an Informed Consent. A survey was conducted using validated questionnaires. Anamnestic data, anthropometry, functional examinations were collected from the all participants. Total antioxidant activity, MDA-oxLDL, SOD, MCP-1, TNF-alpha, IL-1, IL-6, IL-8, total cholesterol, TG, HDL and LDL were determined in blood serum using enzyme immunoassay, multiplex assay and enzymatic method. Results: The analysis of the data obtained revealed a difference in concentrations between the studied subgroups for the total antioxidant activity, SOD, MCP-1 and IL-6. In patients with IHD, the level of the total antioxidant activity was significantly lower by 2.9 times and MCP-1 by 1.4 times, the concentration of SOD and IL-6 was higher by 16 and 30%, respectively. For interpretation, we used a scatter diagram of the density of values - SHAP, reflecting the degree of influence of each parameter on the resulting variable. The greatest contributions to the model project were made by: the total antioxidant activity, SOD, MCP-1, IL-6 and HDL. Conclusions: The data obtained by us can be recommended for the development of a coronary heart disease risk meter in young people, which includes the most significant laboratory parameters. This study was conducted within the framework of the project RSF 21-15-00022.

Financial Risk Meter for emerging markets

In this paper, the daily systemic risk measure FRM (Financial Risk Meter) is proposed for emerging markets (FRM@EM). The FRM@EM is applied to capture systemic risk behavior embedded in the returns of the 25 largest EMs’ Financial Institutions (FIs), covering the BRIMST (Brazil, Russia, India, Mexico, South Africa, and Turkey), and thereby reflects the financial linkages between these economies. The results indicate that the FRM of EMs’ FIs reached its maximum during the US financial crisis following the COVID-19 crisis. In addition, we find that the Macro factors explain the BRIMST's FIs with various degrees of sensibility. Moreover, we propose a robust and well-diversified tail-event and cluster risk-sensitive portfolio allocation model named uplifted hierarchical risk parity (upHRP) and compare it to more classical approaches. Results indicate that the upHRP approach provides better diversification. Moreover, the upHRP portfolio overweights low-central FIs and underweights high-central ones.

Financial Risk Meter FRM based on Expectiles

The Financial Risk Meter (FRM) is an established quantitative tool that, based on conditional Value at Risk (VaR) ideas, yields insight into the dynamics of network risk. Originally, the FRM has been composed via Lasso based quantile regression, but we here extend it by incorporating the idea of expectiles, thus indicating not only the tail probability but rather the actual tail loss given a stress situation in the network. The expectile variant of the FRM enjoys several advantages: Firstly, the multivariate tail risk indicator conditional expectile-based VaR (CoEVaR) can be derived, which is sensitive to the magnitude of extreme losses. Next, FRM index is not restricted to an index compared to the quantile based FRM mechanisms, but can be expanded to a set of systemic tail risk indicators, which provide investors with numerous tools in terms of diverse risk preferences. The power of FRM also lies in displaying the FRM distribution across various entities every day. In a functional data context, the FRM identifies outlying curves and serves as a signal box to display aberrant functional behavior. Two distinct patterns can be discovered under high stress and during stable periods from the empirical results in the United States stock market. Furthermore, the framework is able to identify individual risk characteristics and to capture spillover effects in a network.

Financial Risk Meter Based on Expectiles

The Financial Risk Meter (FRM) is an established quantitative tool that, based on conditional Value at Risk (VaR) ideas, yields insight into the dynamics of network risk. Originally, the FRM has been composed via Lasso based quantile regression, but we here extend it by incorporating the idea of expectiles, thus indicating not only the tail probability but rather the actual tail loss given a stress situation in the network. The expectile variant of the FRM enjoys several advantages: Firstly, the multivariate tail risk indicator conditional expectile-based VaR (CoEVaR) can be derived, which is sensitive to the magnitude of extreme losses. Next, FRM index is not restricted to an index compared to the quantile based FRM mechanisms, but can be expanded to a set of systemic tail risk indicators, which provide investors with numerous tools in terms of diverse risk preferences. The power of FRM also lies in displaying the FRM distribution across various entities every day. Two distinct patterns can be discovered under high stress and during stable periods from the empirical results in the United States stock market. Furthermore, the framework is able to identify individual risk characteristics and to capture spillover effects in a network.

A Financial Risk Meter for China

This paper develops a new risk meter specifically for China – FRM@China – to detect systemic financial risk as well as tail-event (TE) dependencies among major financial institutions (FIs). Compared with the CBOE FIX VIX, which is currently the most popular financial risk measure, FRM@China has less noise. It also emitted a risk signature much earlier than the CBOE FIX VIX index in the 2020 COVID pandemic. In addition, FRM@China uses a single quantile-lasso regression model to allow both the assessment of risk transfer between different sectors in which FIs operate and the prediction of systemic risk. Because the risk indicator in FRM@China is based on penalization terms, its relationship with macro variables are unknown and non-linear. This paper further expands the existing FRM approach by using Shapley values to identify the dynamic contribution of different macro features in this type of "black box" situation. The results show that short-term interest rates and forward guidance are significant risk drivers. This paper considers the interaction among FIs from mainland China, Hong Kong and Taiwan to provide an enhanced regional tool set for regulators to evaluate financial policy responses. All quantlets are available on quantlet.com.

FRM Financial Risk Meter for Emerging Markets

The fast-growing Emerging Market (EM) economies and their improved transparency and liquidity have attracted international investors. However, the external price shocks can result in a higher level of volatility as well as domestic policy instability. Therefore, an efficient risk measure and hedging strategies are needed to help investors protect their investments against this risk. In this paper, a daily systemic risk measure, called FRM (Financial Risk Meter) is proposed. The FRM@ EM is applied to capture systemic risk behavior embedded in the returns of the 25 largest EMs’ FIs, covering the BRIMST (Brazil, Russia, India, Mexico, South Africa, and Turkey), and thereby reflects the financial linkages between these economies. Concerning the Macro factors, in addition to the Adrian & Brunnermeier (2016) Macro, we include the EM sovereign yield spread over respective US Treasuries and the above-mentioned countries’ currencies. The results indicated that the FRM of EMs’ FIs reached its maximum during the US financial crisis following by COVID -9 crisis and the Macro factors explain the BRIMST’ FIs with various degrees of sensibility. We then study the relationship between those factors and the tail event network behavior to build our policy recommendations to help the investors to choose the suitable market for investment and tail-event optimized portfolios. For that purpose, an overlapping region between portfolio optimization strategies and FRM network centrality is developed. We propose a robust and well-diversified tail-event and cluster risk-sensitive portfolio allocation model and compare it to more classical approaches.

Association of polymorphisms of genes TCF7L2, FABP2, KCNQ1, ADIPOQ with the prognosis of the development of type 2 diabetes mellitus

To study the possibility of using polymorphisms of genesTCF7L2,FABP2,KCNQ1,ADIPOQas markers for predicting the development of type 2 diabetes mellitus (T2D) in the population of Novosibirsk. On the basis of prospective observation of a representative population sample of residents of Novosibirsk (HAPIEE), 2 groups were formed according to the case-control principle (case people who had diabetes mellitus 2 over 10 years of observation, and control people who did not developed disorders of carbohydrate metabolism). T2D group (n=443, mean age 56.26.7 years, men 29.6%, women 70.4%), control group (n=532, mean age 56.17.1 years, men 32.7%, women 67.3%). DNA was isolated by phenol-chloroform extraction. Genotyping was performed by the method of polymerase chain reaction with subsequent analysis of restriction fragment length polymorphism, polymerase chain reaction in real time. Statistical processing was carried out using the SPSS 16.0 software package. No significant effect of rs1799883 of theFABP2gene, rs2237892 of theKCNQ1gene, and rs6773957 of theADIPOQgene on the risk of developing T2D was found. Genotypes TT and TC rs7903146 of theTCF7L2gene are genotypes for the risk of developing T2D (relative risk RR 3.90, 95% confidence interval CI 2.316.61,p0.001; RR 1.86, 95% CI 1.422.43,p0.001, respectively). The CC genotype rs7903146 of theTCF7L2gene is associated with a protective effect against T2D (RR 0.37, 95% CI 0.290.49,p0.001). When theTCF7L2gene is included in the model for assessing the risk of developing T2D rs7903146, it retains its significance in both men and women. The rs7903146 polymorphism of theTCF7L2gene confirmed its association with the prognosis of the development of T2D, which indicates the possibility of considering it as a candidate for inclusion in a diabetes risk meter. Variants of risk meters have been developed to assess the prognosis of the development of diabetes mellitus 2 in men and women aged 4569 years during 10 years of follow-up. The association with the prognosis of the development of T2D polymorphisms rs1799883 of theFABP2gene, rs2237892 of theKCNQ1gene and rs6773957 of theADIPOQgene was not found.

Прогнозирование неблагоприятных сердечно-сосудистых событий в постинфарктном периоде с учетом приверженности лечению

Aim: to develop a graphic risk meter scale concerning major adverse cardiovascular events (MCVEs) during the 5-year post-infarction period, taking into account patients’ compliance to the treatment.Patients and Methods: the analyze was conducted on the data from a prospective five-year follow-up of patients (n=115) who experienced a myocardial infarction (MI) and were registered in the database of the Registry of Acute Myocardial Infarction (Tomsk). The following endpoints were analyzed: repeated MI, hospitalization for exacerbation of coronary heart disease, advanced chronic heart failure (CHF), and cardiac arrhythmias.Results: the study identified a set of factors that had a significant impact on the clinical course of the post-infarction period. CHF (diagnosed before the development of index MI) increased the chances of adverse course of post-infarction period in 9.5 times. The lack of achievement of systolic blood pressure target values increased the risk of cardiovascular complications in 5 times. The patients’ age of 75 years and older increased the possibility of an adverse postinfarction period course in 8.5 times. The predictor that favorably affected the post-infarction period course was an integral indicator equal to the product of the specific weight of vital drugs (in the general structure of prescriptions) and the treatment compliance degree. This indicator reduced the risk of MCVEs by 61% for each unit value, i.e. with the highest possible compliance to therapy that meets the recommendations; an adverse course risk of the post-infarction period decreased by 2.5 times.Conclusion: the results emphasize the importance of compliance with medical recommendations by patients who have experienced MI. The proposed graphic risk meter scale can be used to predict the development of adverse MCVEs during the 5-year post-infarction period. KEYWORDS: coronary heart disease, myocardial infarction, cardiovascular risk, treatment compliance, Morisky-Green test, disease prognosis.FOR CITATION: Kuzheleva E.A., Fedyunina V.A., Alexandrenko V.A. et al. Prediction of adverse cardiovascular events in the post-infarction period, taking into account treatment compliance. Russian Medical Inquiry. 2020;4(7):431–436. DOI: 10.32364/2587-6821-2020-4-7-431-436.

AN AI APPROACH TO MEASURING FINANCIAL RISK

AI artificial intelligence brings about new quantitative techniques to assess the state of an economy. Here, we describe a new measure for systemic risk: the Financial Risk Meter (FRM). This measure is based on the penalization parameter (λ) of a linear quantile lasso regression. The FRM is calculated by taking the average of the penalization parameters over the 100 largest US publicly-traded financial institutions. We demonstrate the suitability of this AI-based risk measure by comparing the proposed FRM to other measures for systemic risk, such as VIX, SRISK and Google Trends. We find that mutual Granger causality exists between the FRM and these measures, which indicates the validity of the FRM as a systemic risk measure. The implementation of this project is carried out using parallel computing, the codes are published on www.quantlet.de with keyword [Formula: see text] FRM. The R package RiskAnalytics is another tool with the purpose of integrating and facilitating the research, calculation and analysis methods around the FRM project. The visualization and the up-to-date FRM can be found on hu.berlin/frm.

FRM: A Financial Risk Meter Based on Penalizing Tail Events Occurrence

In this paper we propose a new measure for systemic risk: the Financial Risk Meter (FRM). This measure is based on the penalization parameter (λ) of a linear quantile lasso regression. The FRM is calculated by taking the average of the penalization parameters over the 100 largest US publicly traded financial institutions. We demonstrate the suitability of this risk measure by comparing the proposed FRM to other measures for systemic risk, such as VIX, SRISK and Google Trends. We find that mutual Granger causality exists between the FRM and these measures, which indicates the validity of the FRM as a systemic risk measure. The implementation of this project is carried out using parallel computing, the codes are published on www.quantlet.de with keyword FRM. The R package RiskAnalytics is another tool with the purpose of integrating and facilitating the research, calculation and analysis methods around the FRM project.

Healthcare quality maturity assessment model based on quality drivers.

Purpose - Healthcare providers differ in their readiness and maturity levels regarding quality and quality management systems applications. The purpose of this paper is to serve as a useful quantitative quality maturity-level assessment tool for healthcare organizations. Design/methodology/approach - The model proposes five quality maturity levels (chaotic, primitive, structured, mature and proficient) based on six quality drivers: top management, people, operations, culture, quality focus and accreditation. Findings - Healthcare managers can apply the model to identify the status quo, quality shortcomings and evaluating ongoing progress. Practical implications - The model has been incorporated in an interactive Excel worksheet that visually displays the quality maturity-level risk meter. The tool has been applied successfully to local hospitals. Originality/value - The proposed six quality driver scales appear to measure healthcare provider maturity levels on a single quality meter.

What’s the risk? A comparison of actual and perceived driving risk

It has long been presumed that drivers’ perceptions of risk play an important role in guiding on-road behaviour. The answer to how accurately drivers perceive the momentary risk of a driving situation, however, is unknown. This research compared drivers’ perceptions of the momentary risk for a range of roads to the objective risk associated with those roads. Videos of rural roads, filmed from the drivers’ perspective, were presented to 69 participants seated in a driving simulator while they indicated the momentary levels of risk they were experiencing by moving a risk meter mounted on the steering wheel. Estimates of the objective levels of risk for the roads were calculated using road protection scores from the KiwiRAP database (part of the International Road Assessment Programme). Subsequently, the participants also provided risk estimates for still photos taken from the videos. Another group of 10 participants viewed the videos and photos while their eye movements and fixations were recorded. In a third experiment, 14 participants drove a subset of the roads in a car while providing risk ratings at selected points of interest. Results showed a high degree of consistency across the different methods. Certain road situations were rated as being riskier than the objective risk, and perhaps more importantly, the risk of other situations was significantly under-rated. Horizontal curves and narrow lanes were associated with over-rated risk estimates, while intersections and roadside hazards such as narrow road shoulders, power poles and ditches were significantly under-rated. Analysis of eye movements indicated that drivers did not fixate these features and that the spread of fixations, pupil size and eye blinks were significantly correlated with the risk ratings. An analysis of the road design elements at 77 locations in the video revealed five road characteristics that predicted nearly 80% of the variance in drivers’ risk perceptions; horizontal curvature, lane and shoulder width, gradient, and the presence of median barriers.