Risks, Volume 9, Issue 3 (March 2021) – 13 articles

This work documents a persistent life expectancy heterogeneity by gender and geography in Italy during the period 1995–2019. Based on deviations of life expectancy at age 65, it quantifies the implicit tax/subsidy mechanism triggered when pensions annuities are computed by adopting the same value of longevity for the whole population. The intensity of this transfer mechanism is then measured and projected over the decade 2020–2030. Results show that females are subsidized while males are taxed by around $10\%$. Differences by geography persist along the Italian territory. Since 1995 the macroarea of Mezzogiorno has been taxed by $2\%$, Center and North-West macroareas are being subsidized by around $1\%$, whereas North-East by $2\%$. The intensity of the mechanism, despite decreases over time, is higher among females since the year 2000. From a geographical perspective, the macroarea of Mezzogiorno shows the lowest intensity, but also the lowest reduction as compared to other macroareas. Projections indicate that the North-South divide in this implicit transfer mechanism will persist over the next decade. Full article

This study aims to check market reaction to filing for bankruptcy and restructuring proceedings and to verify the short-term effect of a price reversal in the Polish market in the years 2004–2019. The research was conducted by dividing the analysed companies according to the procedure (bankruptcy and restructuring) and market (the main market and the NewConnect market). The research methodology used in the study is the event analysis method (AR, CAR, AAR and CAAR rates were used in the research), with a few statistical tests (T-test, Generalized rank Z Test, Generalized rank T-Test, Patell or Standardized Residual Test, Kolari and Pynnönen adjusted Patell or Standardized Residual Test). It was found that share prices in the Polish share market react quickly to public information about filing an application for bankruptcy or restructuring. For all analysed companies, the mean rate of return on the event day was equal to −14%, and on the next day, it was −3%. Regardless of the type of share market and the form of proceedings, the reversal effect was not confirmed in the short term. It was found that cumulative above-average rates of return fall more strongly for companies listed on the less liquid Newconnect market (−23.6%), and when information on the filing for bankruptcy proceedings is provided (−28.5%), as opposed to the main market (−19.1%) and restructuring proceedings (−17%). The cumulative average rate of return for all analysed companies in the research period (−2, +10 days) was equal to −20.6%. Full article

The aim of this article is to present the risk model premises related to worker recruitment. Recruitment affects the final selection of workers, whose activities contribute to corporate competitive advantages. Hiring unfavorable workers can influence the results produced by an organization. This risk mostly affects situations when searching for workers via the external labor market, although it can also affect internal recruitment. Therefore, it is necessary to attempt to identify recruitment risk determinants and classify their meaning in such processes. Model formation has both theoretical and intuitive characteristics. Model dependencies and their characteristics are identified in this paper. We attempted to assess the usability of the risk model for economic praxis. The analyses and results provide a model identification of dependencies between the factors determining a workers recruitment process and the risk which is caused by this process (employing inadequate workers who do not meet the employer’s expectations). The identification of worker recruitment process determinants should allow for practically reducing the risk of employing an inadequate worker and contribute to the reduction in unfavorable recruitment processes. The added value of this publication is the complex identification of recruitment process risk determinants and dependency formulations in a model form. Full article

Predicting if a client is worth giving a loan—credit scoring—is one of the most essential and popular problems in banking. Predictive models for this goal are built on the assumption that there is a dependency between the client’s profile before the loan approval and their future behavior. However, circumstances that cause changes in the client’s behavior may not depend on their will and cannot be predicted by their profile. Such clients may be considered “noisy” as their eventual belonging to the defaulters class results rather from random factors than from some predictable rules. Excluding such clients from the dataset may be helpful in building more accurate predictive models. In this paper, we report on primary results on testing the hypothesis that a client can become a defaulter in two scenarios: intentionally and unintentionally. We verify our hypothesis applying data driven regularized classification using an autoencoder to client profiles. To model an intention as a hidden variable, we propose an especially designed regularizer for the autoencoder. The regularizer aims to obtain a representation of defaulters that includes a cluster of intentional defaulters and unintentional defaulters as outliers. The outliers were detected by our model and excluded from the dataset. This improved the credit scoring model and confirmed our hypothesis. Full article

For calculating non-life insurance premiums, actuaries traditionally rely on separate severity and frequency models using covariates to explain the claims loss exposure. In this paper, we focus on the claim severity. First, we build two reference models, a generalized linear model and a generalized additive model, relying on a log-normal distribution of the severity and including the most significant factors. Thereby, we relate the continuous variables to the response in a nonlinear way. In the second step, we tune two random forest models, one for the claim severity and one for the log-transformed claim severity, where the latter requires a transformation of the predicted results. We compare the prediction performance of the different models using the relative error, the root mean squared error and the goodness-of-lift statistics in combination with goodness-of-fit statistics. In our application, we rely on a dataset of a Swiss collision insurance portfolio covering the loss exposure of the period from 2011 to 2015, and including observations from 81 309 settled claims with a total amount of CHF 184 mio. In the analysis, we use the data from 2011 to 2014 for training and from 2015 for testing. Our results indicate that the use of a log-normal transformation of the severity is not leading to performance gains with random forests. However, random forests with a log-normal transformation are the favorite choice for explaining right-skewed claims. Finally, when considering all indicators, we conclude that the generalized additive model has the best overall performance. Full article

The aim of this paper is to merge order statistics with natural catastrophe reinsurance pricing to develop new theoretical and practical insights relevant to market practice and model development. We present a novel framework to quantify the role that occurrence losses (order statistics) play in pricing of catastrophe excess of loss (catXL) contracts. Our framework enables one to analytically quantify the contribution of a given occurrence loss to the mean and covariance structure, before and after the application of a catXL contract. We demonstrate the utility of our framework with an application to idealized catastrophe models for a multi-peril and a hurricane-only case. For the multi-peril case, we show precisely how contributions to so-called lower layers are dominated by high frequency perils, whereas higher layers are dominated by low-frequency high severity perils. Our framework enables market practitioners and model developers to assess and understand the impact of altered model assumptions on the role of occurrence losses in catXL pricing. Full article

Best practice life expectancy has recently been modeled using extreme value theory. In this paper we present the Gumbel autoregressive model of order one—Gumbel AR(1)—as an option for modeling best practice life expectancy. This class of model represents a neat and coherent framework for modeling time series extremes. The Gumbel distribution accounts for the extreme nature of best practice life expectancy, while the AR structure accounts for the temporal dependence in the time series. Model diagnostics and simulation results indicate that these models present a viable alternative to Gaussian AR(1) models when dealing with time series of extremes and merit further exploration. Full article

In micro-lending markets, lack of recorded credit history is a significant impediment to assessing individual borrowers’ creditworthiness and therefore deciding fair interest rates. This research compares various machine learning algorithms on real micro-lending data to test their efficacy at classifying borrowers into various credit categories. We demonstrate that off-the-shelf multi-class classifiers such as random forest algorithms can perform this task very well, using readily available data about customers (such as age, occupation, and location). This presents inexpensive and reliable means to micro-lending institutions around the developing world with which to assess creditworthiness in the absence of credit history or central credit databases. Full article

To build adequate predictive models, a substantial amount of data is desirable. However, when expanding to new or unexplored territories, this required level of information is rarely always available. To build such models, actuaries often have to: procure data from local providers, use limited unsuitable industry and public research, or rely on extrapolations from other better-known markets. Another common pathology when applying machine learning techniques in actuarial domains is the prevalence of imbalanced classes where risk events of interest, such as mortality and fraud, are under-represented in data. In this work, we show how an implicit model using the Generative Adversarial Network (GAN) can alleviate these problems through the generation of adequate quality data from very limited or highly imbalanced samples. We provide an introduction to GANs and how they are used to synthesize data that accurately enhance the data resolution of very infrequent events and improve model robustness. Overall, we show a significant superiority of GANs for boosting predictive models when compared to competing approaches on benchmark data sets. This work offers numerous of contributions to actuaries with applications to inter alia new sample creation, data augmentation, boosting predictive models, anomaly detection, and missing data imputation. Full article

In the context of current crises following COVID-19 and growing global economic uncertainties, the issues regarding financial transactions with FINTECH are increasingly apparent. Consequently, in our opinion, the utilization of FINTECH financial transactions leads to a risk-reduction approach when in contact with other people. Moreover, financial transactions with FINTECH can save up customers’ pecuniary funds. Therefore, during crises, FINTECH applications can be perceived as more competitive than the traditional banking system. All the above have provoked us to conduct research related to the utilization of financial transactions with FINTECH before and after the COVID-19 crisis outbreak. The aim of the article is to present a survey analysis of FINTECH utilization of individual customers before and after the crisis in Bulgaria. The methodology includes a questionnaire survey of 242 individual respondents. For the data processing, we implemented statistical measures and quantitative methods, including two-sample paired t-tests, Levene’s test, and ANOVAs performed through the computer language Python in a web-based interactive computing environment for creating documents, Jupyter Notebook. The findings bring out the main issues related to the implementation of financial transactions with FINTECH under the conditions of the crisis. The findings include the identification of problems related to FINTECH transactions during the COVID-19 crisis in Bulgaria. Full article

Death benefits are generally the largest cash flow items that affect the financial statements of life insurers; some may still not have a systematic process to track and monitor death claims. In this article, we explore data clustering to examine and understand how actual death claims differ from what is expected—an early stage of developing a monitoring system crucial for risk management. We extended the k-prototype clustering algorithm to draw inferences from a life insurance dataset using only the insured’s characteristics and policy information without regard to known mortality. This clustering has the feature of efficiently handling categorical, numerical, and spatial attributes. Using gap statistics, the optimal clusters obtained from the algorithm are then used to compare actual to expected death claims experience of the life insurance portfolio. Our empirical data contained observations of approximately 1.14 million policies with a total insured amount of over 650 billion dollars. For this portfolio, the algorithm produced three natural clusters, with each cluster having lower actual to expected death claims but with differing variability. The analytical results provide management a process to identify policyholders’ attributes that dominate significant mortality deviations, and thereby enhance decision making for taking necessary actions. Full article

The way one leads their life is considered an important factor in health. In this paper we propose a system to provide risk assessment based on behavior for the health insurance sector. To do so we built a platform to collect real-world data that enumerate different aspects of behavior, and a simulator to augment actual data with synthetic. Using the data, we built classifiers to predict variations in important quantities for the lifestyle of a person. We offer a risk assessment service to the health insurance professionals by manipulating the classifier predictions in the long-term. We also address virtual coaching by using explainable Artificial Intelligence (AI) techniques on the classifier itself to gain insights on the advice to be offered to insurance customers. Full article

We introduce four variants of the common age effect model proposed by Kleinow, which describes the mortality rates of multiple populations. Our model extensions are based on the assumption of multiple common age effects, each of which is shared only by a subgroup of all considered populations. This makes the models more realistic while still keeping them as parsimonious as possible, improving the goodness of fit. We apply different clustering methods to identify suitable subgroups. Some of the algorithms are borrowed from the unsupervised learning literature, while others are more domain-specific. In particular, we propose and investigate a new model with fuzzy clustering, in which each population’s individual age effect is a linear combination of a small number of age effects. Due to their good interpretability, our clustering-based models also allow some insights in the historical mortality dynamics of the populations. Numerical results and graphical illustrations of the considered models and their performance in-sample as well as out-of-sample are provided. Full article