Risks, Volume 11, Issue 9 (September 2023) – 15 articles

The purpose and novelty of this article is to investigate the extent to which artificial intelligence chatbot ChatGPT can grasp concepts from quantitative risk management. To this end, we enter a scholarly discussion with ChatGPT in the form of questions and answers, and analyze the responses. The questions are classics from undergraduate and graduate courses on quantitative risk management, and address risk in general, risk measures, time series, extremes and dependence. As a result, the non-technical aspects of risk (such as explanations of various types of financial risk, the driving factors underlying the financial crisis of 2007 to 2009, or a basic introduction to the Basel Framework) are well understood by ChatGPT. More technical aspects (such as mathematical facts), however, are often inaccurate or wrong, partly in rather subtle ways not obvious without expert knowledge, which we point out. The article concludes by providing guidance on the types of applications for which consulting ChatGPT can be useful in order to enhance one’s own knowledge of quantitative risk management (e.g., using ChatGPT as an educational tool to test one’s own understanding of an already grasped concept, or using ChatGPT as a practical tool for identifying risks just not on one’s own radar), and points out those applications for which the current version of ChatGPT should not be invoked (e.g., for learning mathematical concepts, or for learning entirely new concepts for which one has no basis of comparison to assess ChatGPT’s capabilities). Full article

Financial technologies (fintechs) are continuously expanding, across different markets and financial services. While financial technologies bring many opportunities, such as reduced costs and extended inclusion, they also bring risks, among which include cyber risks, that are difficult to measure. One of the difficulties that arise in the measurement of cyber risks is the interdependence among cyber losses, a problem that has not yet been solved. To fill the gap, this paper proposes a multivariate model for cyber risks, based on their observed time series of counts. The time-varying intensity parameter of the model determines the probability that a cyber attack occurs, and its specification takes not only time but also sectorial interdependence into account. The effectiveness of the proposed model is demonstrated by means of a real cyber loss dataset, in which there exists time and sectorial dependence among different events. Full article

Accurate forecasting of insurance claims is of the utmost importance for insurance activity as the evolution of claims determines cash outflows and the pricing, and thus the profitability, of the underlying insurance coverage. These are used as inputs when the insurance company drafts its business plan and determines its risk appetite, and the respective solvency capital required (by the regulators) to absorb the assumed risks. The conventional claim forecasting methods attempt to fit (each of) the claims frequency and severity with a known probability distribution function and use it to project future claims. This study offers a fresh approach in insurance claims forecasting. First, we introduce two novel sets of variables, i.e., weather conditions and car sales, and second, we employ a battery of Machine Learning (ML) algorithms (Support Vector Machines—SVM, Decision Trees, Random Forests, and Boosting) to forecast the average (mean) insurance claim per insured car per quarter. Finally, we identify the variables that are the most influential in forecasting insurance claims. Our dataset comes from the motor portfolio of an insurance company operating in Athens, Greece and spans a period from 2008 to 2020. We found evidence that the three most informative variables pertain to the new car sales with a 3-quarter and 1-quarter lag and the minimum temperature of Elefsina (one of the weather stations in Athens) with a 3-quarter lag. Among the models tested, Random Forest with limited depth and XGboost run on the 15 most informative variables, and these exhibited the best performance. These findings can be useful in the hands of insurers as they can consider the weather conditions and the new car sales among the parameters that are considered to perform claims forecasting. Full article

Modelling claim frequency and claim severity are topics of great interest in property-casualty insurance for supporting underwriting, ratemaking, and reserving actuarial decisions. Standard Generalized Linear Models (GLM) frequency–severity models assume a linear relationship between a function of the response variable and the predictors, independence between the claim frequency and severity, and assign full credibility to the data. To overcome some of these restrictions, this paper investigates the predictive performance of Gradient Boosting with decision trees as base learners to model the claim frequency and the claim severity distributions of an auto insurance big dataset and compare it with that obtained using a standard GLM model. The out-of-sample performance measure results show that the predictive performance of the Gradient Boosting Model (GBM) is superior to the standard GLM model in the Poisson claim frequency model. Differently, in the claim severity model, the classical GLM outperformed the Gradient Boosting Model. The findings suggest that gradient boost models can capture the non-linear relation between the response variable and feature variables and their complex interactions and thus are a valuable tool for the insurer in feature engineering and the development of a data-driven approach to risk management and insurance. Full article

In this paper, we consider the problem of pricing variance, volatility, covariance and correlation swaps for financial markets with semi-Markov volatilities. The paper’s motivation derives from the fact that in many financial markets, the inter-arrival times between book events are not independent or exponentially distributed but instead have an arbitrary distribution, which means they can be accurately modelled using a semi-Markov process. Through the results of the paper, we hope to answer the following question: Is it possible to calculate averaged swap prices for financial markets with semi-Markov volatilities? This question has not been considered in the existing literature, which makes the paper’s results novel and significant, especially when one considers the increasing popularity of derivative securities such as swaps, futures and options written on the volatility index VIX. Within this paper, we model financial markets featuring semi-Markov volatilities and price-averaged variance, volatility, covariance and correlation swaps for these markets. Formulas used for the numerical evaluation of averaged variance, volatility, covariance and correlation swaps with semi-Markov volatilities are presented as well. The formulas that are detailed within the paper are innovative because they provide a new, simplified method to price averaged swaps, which has not been presented in the existing literature. A numerical example involving the pricing of averaged variance, volatility, covariance and correlation swaps in a market with a two-state semi-Markov process is presented, providing a detailed overview of how the model developed in the paper can be used with real-life data. The novelty of the paper lies in the closed-form formulas provided for the pricing of variance, volatility, covariance and correlation swaps with semi-Markov volatilities, as they can be directly applied by derivative practitioners and others in the financial industry to price variance, volatility, covariance and correlation swaps. Full article

In this article, we study stochastic orders over an interval. Mainly, we focus on orders related to the Laplace transform. The results are then applied to obtain a bound for heavy-tailed distributions and are illustrated by some examples. We also indicate how these ordering relationships can be adapted to the classical risk model in order to derive a moment bound for ruin probability. Finally, we compare it with other existing bounds. Full article

Healthcare fraud is intentionally submitting false claims or producing misinterpretation of facts to obtain entitlement payments. Thus, it wastes healthcare financial resources and increases healthcare costs. Subsequently, fraud poses a substantial financial challenge. Therefore, supervised machine and deep learning analytics such as random forest, logistic regression, and artificial neural networks are successfully used to detect healthcare insurance fraud. This study aims to develop a health model that automatically detects fraud from health insurance claims in Saudi Arabia. The model indicates the greatest contributing factor to fraud with optimal accuracy. The labeled imbalanced dataset used three supervised deep and machine learning methods. The dataset was obtained from three healthcare providers in Saudi Arabia. The applied models were random forest, logistic regression, and artificial neural networks. The SMOT technique was used to balance the dataset. Boruta object feature selection was applied to exclude insignificant features. Validation metrics were accuracy, precision, recall, specificity, F1 score, and area under the curve (AUC). Random forest classifiers indicated policy type, education, and age as the most significant features with an accuracy of 98.21%, 98.08% precision, 100% recall, an F1 score of 99.03%, specificity of 80%, and an AUC of 90.00%. Logistic regression resulted in an accuracy of 80.36%, 97.62% precision, 80.39% recall, an F1 score of 88.17%, specificity of 80%, and an AUC of 80.20%. ANN revealed an accuracy of 94.64%, 98.00% precision, 96.08% recall, an F1 score of 97.03%, a specificity of 80%, and an AUC of 88.04%. This predictive analytics study applied three successful models, each of which yielded acceptable accuracy and validation metrics; however, further research on a larger dataset is advised. Full article

This study demonstrates the significant impact of market sentiment, derived from social media, on the daily price prediction of cryptocurrencies in both bull and bear markets. Through the analysis of approximately 567 thousand tweets related to twelve specific cryptocurrencies, we incorporate the sentiment extracted from these tweets along with daily price data into our prediction models. We test various algorithms, including ordinary least squares regression, long short-term memory network and neural hierarchical interpolation for time series forecasting (NHITS). All models show better performance once the sentiment is incorporated into the training data. Beyond merely assessing prediction error, we scrutinise the model performances in a practical setting by applying them to a basic trading algorithm managing three distinct portfolios: established tokens, emerging tokens, and meme tokens. While NHITS emerged as the top-performing model in terms of prediction error, its ability to generate returns is not as compelling. Full article

This study assesses the dynamic relationship between macroeconomic factors and bank asset quality based on changes in the condition of stock market returns. A dynamic panel two-step system, the Generalized Method of Moments (system GMM) model, is employed using panel data from 18 universal banks spanning the period of 2007 to 2021. The analysis revealed that the real GDP growth rate, the average lending rate, and the real exchange rate represent a set of macroeconomic factors with a marked influence on banks’ asset quality, where a unit increase in these variables drive 0.02 percent, 0.98 percent, and 0.27 percent improvement in asset quality, respectively. In addition, a high-inflation rate was found to exert an adverse effect of −0.32 percent on asset quality, as it affects borrowers’ financial ability to meet loan repayment obligations. Furthermore, the study verified the existence of a positive relationship between market condition and asset quality, where a rise in the market return drives a 0.07 percent improvement in bank asset quality. This implies that bank performance adapts to changes in market conditions as posited under the Adaptive Market Hypothesis (AMH). Bank managers should consolidate banks’ asset bases during conditions of market stability to withstand periodic market fluctuations to boost trading momentum. Policy recommendations are suggested to foster a conducive business environment for bank stability. Full article

The goal is to create a systemic risk profile of companies during the COVID-19 crisis, which reflects their cause-and-effect relationships and risk management. The research objects are the following types of risks for companies listed in “Global-500” (Fortune) and the top 55 most competitive digital economies of the world (IMD) in 2017–2022: (1) risk of reduction in competitiveness (rank), (2) risk of reduction in revenue, and (3) risk of reduction in profit. The research methodology is based on the method of structural equation modelling (SEM), which allowed for exploring the cause-and-effect relationships between risk changes and digital risk management for companies during the COVID-19 crisis. As a result, based on the SEM model, it was proven that risks for companies during the COVID-19 crisis only slightly increased compared with that at the pre-crisis level. It was determined that companies faced large risks during the COVID-19 crisis in developed countries. It was discovered that, due to successful adaptation, risk management of companies assuaged the manifestations of the COVID-19 crisis in the economy. The key conclusion is that, under the conditions of a crisis of a non-economic nature (e.g., the COVID-19 crisis), companies independently and successfully manage their risks with the help of measures of digitalisation: corporate risk management with the limitation of state intervention is preferable. The contribution to the literature consists of the development of the concept of risks for companies by clarifying the specifics of risks and risk management of companies during the COVID-19 crisis. The theoretical significance lies in the fact that the authors’ conclusions rethought the risks for companies under the conditions of a crisis given the special context of a crisis of a non-economic nature (via the example of the COVID-19 crisis). The practical significance is that the developed novel approach to risk management of companies through digitalisation, which is based on the experience of the COVID-19 crisis, will be useful for risk management of companies under the conditions of future crises of non-economic nature caused by epidemics/pandemics and/or environmental disasters. Full article

Constructing an accurate model for insurance losses is a challenging task. Researchers have developed various methods to model insurance losses, such as composite models. Composite models combine two distributions: one for part of the data with small and high frequencies and the other for large values with low frequencies. The purpose of this article is to consider a mixture of prior distributions for exponential–Pareto and inverse-gamma–Pareto composite models. The general formulas for the posterior distribution and the Bayes estimator of the support parameter $\theta$ are derived. It is shown that the posterior distribution is a mixture of individual posterior distributions. Analytic results and Bayesian inference based on the proposed mixture prior distribution approach are provided. Simulation studies reveal that the Bayes estimator with a mixture distribution outperforms the Bayes estimator without a mixture distribution and the ML estimator regarding their accuracies. Based on the proposed method, the insurance losses from natural events, such as floods from 2000 to 2019 in the USA, are considered. As a measure of goodness-of-fit, the Bayes factor is used to choose the best-fitted model. Full article

Pandemic bonds can be used as an effective tool to mitigate the economic losses that governments face during pandemics and transfer them to the global capital market. Once considered as an “uninsurable” event, pandemic bonds caught the attention of the world with the issuance of pandemic bonds by the World Bank in 2017. Compared to other CAT bonds, pandemic bonds received less attention from actuaries, industry professionals, and academic researchers. Existing research focused mainly on how to bring epidemiological parameters to the pricing mechanism through compartmental models. In this study, we introduce the stochastic logistic growth model-based pandemic bond pricing framework. We demonstrate the proposed model with two numerical examples. First, we calculate what investor is willing to pay for the World Bank issued pandemic bond while accounting for possible future pandemic, but require to have the same yield to maturity when no pandemic is there, and without using COVID-19 data. In the second example, we calculate the fair value of a pandemic bond with characteristics similar to the World Bank issued pandemic bond, but using COVID-19 data. The model can be used as an alternative to epidemic compartmental model-based pandemic bond pricing mechanisms. Full article

Cyber security importance has escalated globally, driven by its pivotal role in shaping daily life, encompassing both personal and non-personal aspects. Cyber security breach probability functions play a crucial role in comprehending how cyber security investments affect vulnerability to cyber attacks. These functions employ mathematical models to guide decision making in cyber risk management. Thus, studying and improving them is useful in this context. In particular, using these models, this article explores the effectiveness of an integrated risk management strategy that merges insurance and security investments, aiming to minimize overall security expenses. Within this strategy, security investments contribute to reducing the insurance premium. This research investigates the optimal investment for this blended approach under total insurance coverage. When the integrated risk management strategy combining insurance and security investments is deemed the optimal choice, this paper reveals that the insurance premium tends to be the dominant component in the overall security expense in the majority of cases. This implies that the cost of insurance outweighs the cost of security investments. Full article

In light of the function of Internal Auditing and its significance in assessing and ensuring the validity of data, information, reports, and high lists generated by the Accounting Information System and improving its credibility and dependability, the purpose of this study was to investigate the relationship between Internal Auditing (IA) and Accounting Information System (AIS) in Jordanian banks, with a focus on the mediator role of Organizational Commitment (OC). A cross-sectional survey method was used to collect data from a sample of employees who work in banks, including those who work in the internal audit department. The collected data were analyzed using SPSS 26.0 and PROCESS V4.1. The study sample includes 193 employees who work in banks, including those who work in the internal audit department. Descriptive statistical methods, such as frequencies, percentages, means, and standard deviations, were employed to depict both the characteristics of the sample and the participants’ responses to the study items. The results indicate that IA has a positive relationship with AIS. Moreover, the results indicate that OC partially mediates the relationship between IA and AIS in Jordanian banks. Full article

We apply a Markov-switching Bayesian vector autoregression (MSBVAR) model to mortality forecasting. MSBVAR has not previously been applied in this context, and our results show that it is a promising tool for mortality forecasting. Our model shows better forecasting accuracy than the Lee–Carter and Bayesian vector autoregressive (BVAR) models without regime-switching and while retaining the advantages of BVAR. MSBVAR provides more reliable estimates for parameter uncertainty and more flexibility in the shapes of point-forecast curves and shapes of confidence intervals than BVAR. Through regime-switching, MSBVAR helps to capture transitory changes in mortality and provides insightful quantitative information about mortality dynamics. Full article