Showing papers on "Credit risk published in 2019"

Machine Learning in Banking Risk Management: A Literature Review

Abstract: There is an increasing influence of machine learning in business applications, with many solutions already implemented and many more being explored. Since the global financial crisis, risk management in banks has gained more prominence, and there has been a constant focus around how risks are being detected, measured, reported and managed. Considerable research in academia and industry has focused on the developments in banking and risk management and the current and emerging challenges. This paper, through a review of the available literature seeks to analyse and evaluate machine-learning techniques that have been researched in the context of banking risk management, and to identify areas or problems in risk management that have been inadequately explored and are potential areas for further research. The review has shown that the application of machine learning in the management of banking risks such as credit risk, market risk, operational risk and liquidity risk has been explored; however, it doesn’t appear commensurate with the current industry level of focus on both risk management and machine learning. A large number of areas remain in bank risk management that could significantly benefit from the study of how machine learning can be applied to address specific problems.

Two-stage consumer credit risk modelling using heterogeneous ensemble learning

Abstract: Modelling consumer credit risk is a crucial task for banks and non-bank financial institutions to support decision-making on granting loans. To model the overall credit risk of a consumer loan in terms of expected loss (EL), three key credit risk parameters must be estimated: probability of default (PD), loss given default (LGD) and exposure at default (EAD). Research to date has tended to model these parameters separately. Moreover, a neglected area in the field of LGD/EAD modelling is the application of ensemble learning, which by benefitting from diverse base learners reduces the over-fitting problem and enables modelling diverse risk profiles of defaulted loans. To overcome these problems, this paper proposes a two-stage credit risk model that integrates (1) class-imbalanced ensemble learning for predicting PD (credit scoring), and (2) an EAD prediction using a regression ensemble. Furthermore, multi-objective evolutionary feature selection is used to minimize both the misclassification cost (root mean squared error) of the PD and EAD models and the number of attributes necessary for modelling. For this task, we propose a misclassification cost metric suitable for consumer loans with fixed exposure because it combines opportunity cost and LGD. We show that the proposed credit risk model is not only more effective than single-stage credit risk models but also outperforms state-of-the-art methods used to model credit risk in terms of prediction and economic performance.

Solvency: Models, Assessment and Regulation

Abstract: Introduction General Outline of the Book Organizations A Selection of Solvency Readings Solvency: What Is It? In the 18th Century What Does Solvency Mean? PAST AND PRESENT: A HISTORICAL REVIEW AND DIFFERENT APPROACHES TO SOLVENCY The European Union: Solvency 0 and Accounting The Works of Campagne Other Steps toward the First Directives The Non-Life Directives (First, Second, and Third) The Life Directives (First, Second, and Third) Calculating the Solvency Margin for Non-Life Insurance Business The Insurance Accounting Directive (IAD) The European Union: Solvency I The Muller Report Comments from Groupe Consultatif The Solvency I Directives Calculating the Solvency Margin for Non-Life Insurance Business Steps toward Solvency II: Bank for International Settlements (BIS): The New Basel Capital Accord IASB: Toward a New Accounting System IAIS: Insurance Principles and Guidelines IAA: A Global Framework for Solvency Assessment EU: Solvency II - Phase I Steps toward Solvency II: 2 Australia Canada Denmark Finland The Netherlands Singapore Sweden Switzerland U.K. U.S. Some Other Systems Summary of Different Systems PRESENT: MODELING A STANDARD APPROACH The Fundamental Ideas A Model for the Solvency Assessment Level of Capital Requirements Risks and Diversification Risk Measures Valuations Fair Value: Introduction Purposes of Valuation Best Estimate of Insurance Liability and Technical Provisions Fair Value Dependencies, Baseline, and Benchmark Models Risk Measures Assume Normality Assume Nonnormality Correlations between Risks: Different Levels of Conservatism Parameters in a Factor-Based Model One Example of Risk Categories and Diversification Insurance Risk Market Risk Credit Risk Operational Risk Liquidity Risk Dependency A Proposal for a Standard Approach: From Formula to Spreadsheet The Insurance Risk, CIR Market Risk, CMR Credit Risk, CCR Operational Risk, COR The Total Factor-Based Model A Spreadsheet Approach Parameter Estimates An Example PART C PRESENT AND FUTURE: EU SOLVENCY II - PHASE 2: GROUPS AND INTERNAL MODELING IN BRIEF The European Union: Reinsurance, Insurance Groups, and Financial Conglomerates Reinsurance Insurance Groups and Financial Conglomerates The European Union: Solvency II - Phase II Recommendations for the First Pillar Recommendations for the Second Pillar Recommendations for the Third Pillar General Considerations The First Wave of Requests (Pillar II) The Second Wave of Requests Will Include the Following Issues (Pillar I) The Third Wave of Requests Will Include the Following Issues (Pillar III) A Brief Summary Further Steps Internal Models and Risk Management Forecasting the Future and Risk Management APPENDICES Appendix A A Proposal for a Standard Approach: One Step toward Application Insurance Risk Market Risk Credit Risk Operational Risk Appendix B Insurance Classes Non-Life Classes Life Classes Appendix C From the Non-Life Directives Solvency 0 Solvency I Appendix D From the Life Directives Solvency 0 Solvency I Appendix E IAIS: Insurance Principles, Standards, and Guidelines Principles Standards Guidances Appendix F From the Proposed Reinsurance Directive Chapter 3: Rules Relating to the Solvency Margin and to the Guarantee Fund Appendix G Annex I and Annex II in the Insurance Group Directive Annex I : Calculation of the Adjusted Solvency of Insurance Undertakings Annex II: Supplementary Supervision for Insurance Undertakings That Are Subsidiaries of an Insurance Holding Company, a Reinsurance Undertaking or a Non-Member-Country Insurance Undertaking Appendix H From the Financial Conglomerates Directive Amendments to the Non-Life Directive Made (EEC, 1973) Amendments to the Life Directive (EEC, 1979) Amendments to the Insurance Group Directive (COM, 1998) Annex I: Capital Adequacy Appendix I Prudent Person Rule Article 18: Investment Rules

An Empirical Comparison of Machine-Learning Methods on Bank Client Credit Assessments

Abstract: Machine learning and artificial intelligence have achieved a human-level performance in many application domains, including image classification, speech recognition and machine translation. However, in the financial domain expert-based credit risk models have still been dominating. Establishing meaningful benchmark and comparisons on machine-learning approaches and human expert-based models is a prerequisite in further introducing novel methods. Therefore, our main goal in this study is to establish a new benchmark using real consumer data and to provide machine-learning approaches that can serve as a baseline on this benchmark. We performed an extensive comparison between the machine-learning approaches and a human expert-based model—FICO credit scoring system—by using a Survey of Consumer Finances (SCF) data. As the SCF data is non-synthetic and consists of a large number of real variables, we applied two variable-selection methods: the first method used hypothesis tests, correlation and random forest-based feature importance measures and the second method was only a random forest-based new approach (NAP), to select the best representative features for effective modelling and to compare them. We then built regression models based on various machine-learning algorithms ranging from logistic regression and support vector machines to an ensemble of gradient boosted trees and deep neural networks. Our results demonstrated that if lending institutions in the 2001s had used their own credit scoring model constructed by machine-learning methods explored in this study, their expected credit losses would have been lower, and they would be more sustainable. In addition, the deep neural networks and XGBoost algorithms trained on the subset selected by NAP achieve the highest area under the curve (AUC) and accuracy, respectively.

Machine Learning and AI for Risk Management

Abstract: We explore how machine learning and artificial intelligence (AI) solutions are transforming risk management. A non-technical overview is first given of the main machine learning and AI techniques of benefit to risk management. Then a review is provided, using current practice and empirical evidence, of the application of these techniques to the risk management fields of credit risk, market risk, operational risk, and compliance (‘RegTech’). We conclude with some thoughts on current limitations and views on how the field is likely to develop in the short- to medium-term. Overall, we present an optimistic picture of the role of machine learning and AI in risk management, but note some practical limitations around suitable data management policies, transparency, and lack of necessary skillsets within firms.

A novel ensemble classification model based on neural networks and a classifier optimisation technique for imbalanced credit risk evaluation

Abstract: Significant research has been performed on credit risk evaluation, with many machine learning and data mining techniques being employed for financial decision-making. The back propagation (BP) neural network has been a popular choice for credit risk evaluation problems, but many studies have found classifier ensembles to be superior to single classifiers. In this paper, a novel ensemble model based on the synthetic minority over-sampling technique (SMOTE) and a classifier optimisation technique is proposed for personal credit risk evaluation. To mitigate the negative effects of imbalanced datasets on the performance of the credit evaluation model, the SMOTE technique is used to rebalance the target training dataset. The particle swarm optimisation (PSO) algorithm is employed to search for the best-connected weights and deviations in the BP neural networks. Based on the optimised BP neural network classifiers, an ensemble model is developed that combines the AdaBoost approach with the base classifiers. To ensure that the proposed model provides accurate and stable performance, we thoroughly explore and discuss the optimal parameters for the ensemble classification model. Finally, the proposed ensemble model is tested on German and Australian real-world imbalanced datasets. The results demonstrate that this model is more effective at processing credit data problems compared to the other classification models examined in this study.

The impact of bank capital, bank liquidity and credit risk on profitability in postcrisis period: A comparative study of US and Asia

Abstract: The purpose of this study is to explore the influence of bank capital, bank liquidity level and credit risk on the profitability of commercial banks in the postcrisis period between 2011 and 2017 i...

A dynamic credit risk assessment model with data mining techniques: evidence from Iranian banks

Abstract: Giving loans and issuing credit cards are two of the main concerns of banks in that they include the risks of non-payment. According to the Basel 2 guidelines, banks need to develop their own credit risk assessment systems. Some banks have such systems; nevertheless they have lost a large amount of money simply because the models they used failed to accurately predict customers’ defaults. Traditionally, banks have used static models with demographic or static factors to model credit risk patterns. However, economic factors are not independent of political fluctuations, and as the political environment changes, the economic environment evolves with it. This has been especially evident in Iran after the 2008–2016 USA sanctions, as many previously reliable customers became unable to repay their debt (i.e., became bad customers). Nevertheless, a dynamic model that can accommodate fluctuating politico-economic factors has never been developed. In this paper, we propose a model that can accommodate factors associated with politico-economic crises. Human judgement is removed from the customer evaluation process. We used a fuzzy inference system to create a rule base using a set of uncertainty predictors. First, we train an adaptive network-based fuzzy inference system (ANFIS) using monthly data from a customer profile dataset. Then, using the newly defined factors and their underlying rules, a second round of assessment begins in a fuzzy inference system. Thus, we present a model that is both more flexible to politico-economic factors and can yield results that are max compatible with real-life situations. Comparison between the prediction made by proposed model and a real non-performing loan indicates little difference between them. Credit risk specialists also approve the results. The major innovation of this research is producing a table of bad customers on a monthly basis and creating a dynamic model based on the table. The latest created model is used for assessing customers henceforth, so the whole process of customer assessment need not be repeated. We assert that this model is a good substitute for the static models currently in use as it can outperform traditional models, especially in the face of economic crisis.

The CDS-bond basis

Abstract: We investigate the cross‐sectional variation in the credit default swap (CDS)‐bond bases and test explanations for the violation of the arbitrage relation between cash bond and CDS contract, which states that the basis should be zero in normal conditions. The evidence is consistent with “limits to arbitrage” theories in that deviations are larger for bonds with higher frictions as measured by trading liquidity, funding cost, counterparty risk, and collateral quality. Surprisingly, we find the basis to be more negative when bond lending fee is higher suggesting that arbitrageurs are unwilling to engage in a negative basis trade when short interest on the bond is high.

A Big Data Mining Approach of PSO-Based BP Neural Network for Financial Risk Management With IoT

Abstract: In recent years, the technology about IoT (Internet of Things) has been applied into finance domain, and the generated data, such as the real-time data of chattel mortgage supervision with GPS, sensors, network cameras, mobile devices, etc., has been used to improve the capability of financial credit risk management of bank loans. Financial credit risk is by far one of the most significant risks that commercial banks have to face, however, when confronting to the massively growing financial data from multiple sources including Internet, mobile networks or IoT, traditional statistical models and neural network models might not operate fairly or accurately enough for credit risk assessment with those diverse data. Hence, there is a practical need to establish more powerful risk prediction models with artificial intelligence based on big data analytics to predict default behaviors with better accuracy and capacity. In this article, a big data mining approach of Particle Swarm Optimization (PSO) based Backpropagation (BP) neural network is proposed for financial risk management in commercial banks with IoT deployment, which constructs a nonlinear parallel optimization model with Apache Spark and Hadoop HDFS techniques on the dataset of on-balance sheet item and off-balance sheet item. The experiment results indicate that this parallel risk management model has fast convergence rate and powerful predictive capacity, and performs efficiently in screening default behaviors. In the meanwhile, the distributed implementation on big data clusters largely reduces the processing time of model training and testing.

Credit risk modeling using Bayesian network with a latent variable

Abstract: Credit risk assessment is an important task for the implementation of the bank policies and commercial strategies. In this paper, we used a discrete Bayesian network with a latent variable to model the payment default of loans subscribers. The proposed Bayesian network includes a built-in clustering feature. A full procedure for learning its parameters, based on a customized Expectation-Maximization algorithm was provided. This model allows evaluating the payment default probability taking into account several factors and handling a multi-class situation. Relying on a real data set describing loans contracts, we calibrated the model and performed several analyses. The obtained results highlight a regime switching of the default probability distribution: Two classes were determined showing a change in credit risk profiles.

Blockchains, Real-time Accounting, and the Future of Credit Risk Modeling

Abstract: In this paper I discuss how blockchains potentially could affect the way credit risk is modeled, and how the improved trust and timing associated with blockchain-enabled real-time accounting could improve default prediction. To demonstrate the (quite substantial) effect the change would have on well-known credit risk measures, a simple case-study compares Z-scores and Merton distances to default computed using typical accounting data of today to the same risk measures computed under a hypothetical future blockchain regime.

Limited Investment Capital and Credit Spreads

Abstract: Using proprietary credit default swap (CDS) data, I investigate how capital shocks at protection sellers impact pricing in the CDS market. Seller capital shocks—measured as CDS portfolio margin payments—account for 12% of the time‐series variation in weekly spread changes, a significant amount given that standard credit factors account for 18% during my sample. In addition, seller shocks possess information for spreads that is independent of institution‐wide measures of constraints. These findings imply a high degree of market segmentation, and suggest that frictions within specialized financial institutions prevent capital from flowing into the market at shorter horizons.

The internal determinants of bank profitability and stability: An insight from banking sector of Pakistan

Abstract: The purpose of this study is to examine the internal determinants of bank profitability and stability in Pakistan banking sector. Because of specific research objectives, this study excludes the external factors of profitability and stability to find the role of bank internal determinants in achieving high performance.,A panel regression analysis is built on a balanced panel data using 24 commercial banks over the sample period of 2007-2015. The authors performed a separate analysis of bank profitability and stability. Both models used a comprehensive set of bank internal determinants.,The results that were obtained from profitability model indicated that bank size, credit risk, funding risk and stability have statistically significant impacts on profitability, while liquidity risk showed the statistically insignificant impact on profitability. Regression findings from stability model reveal that bank size, liquidity risk, funding risk and profitability have statistically significant impacts on stability, while credit risk had an insignificant effect on stability. However, the effect of the financial crisis is uniform and showed statistically insignificant impact in both models.,Overall, the authors’ findings bring some new but useful insights to the banking literature. Some recommendations may be functional for the sustainable performance of banks.,In view of study results, the authors provide interesting insights into the practices and characteristics of banks in Pakistan. This study also highlights significant bank internal determinants to improve understanding in the existing literature.

Developments in Risk Management in Islamic Finance: A Review

Abstract: The purpose of this study is to review recent developments pertaining to risk management in Islamic banking and finance literature. The study explores the fundamental features of risks associated with Islamic banks (IBs) as compared to those associated with conventional banks (CBs) in order to determine the extent to which IBs engage in effective risk mitigation. The study includes a consideration of the major studies in which the fundamental features of Islamic banks and finance (IBF) and the main characteristics of risk management in IBs are analyzed in comparison with those of CBs. Specifically, these two kinds of banks are compared in relation to the types of risks faced, the characteristics of those risks, and the nature and extent of exposure to those risks. A tabular methodology approach is used in concert with a comparative literature review approach for the analysis. The results show that there is weak support for Shariah-based product development due to the lack of risk mitigation expertise in IBs. The conclusion presented is that in comparison with CBs, IBs are more risk-sensitive due to the nature of their products, contract structure, legal costing, governance practices, and liquidity infrastructure. Furthermore, the determinants of the credit risk of Islamic banks in Malaysia (MIBs) are examined. Overall, bank capital and financing expansion have a significant negative impact on the credit risk level of IBs in Malaysia.

Financial credit risk prediction in internet finance driven by machine learning

Abstract: The development of science and technology promotes the constant changes of consumer finance, but also brings some financial credit risks. In particular, with the continuous development of Internet finance, financial credit risk is increasingly difficult to control. Based on machine learning algorithm, this study improved the machine learning algorithm and named it MLIA algorithm. Meanwhile, this study decomposes the objective function into weighted sums of several basis functions. This study uses three typical test functions to compare the performance of MLIA prediction algorithm and logistic prediction algorithm. Simultaneously, this study analyzes the performance of MLIA financial credit risk prediction model by taking the data of an Internet financial company as an example. In addition, this study used the AUC (area under curve) value as a specific indicator of model performance verification. Research shows that machine learning has a good predictive effect on MLIA financial credit risk prediction and can provide theoretical reference for subsequent related research.

Time-Varying Asset Volatility and the Credit Spread Puzzle

Abstract: Most extant structural credit risk models underestimate credit spreads—a shortcoming known as the credit spread puzzle. We consider a model with priced stochastic asset risk that is able to fit medium‐ to long‐term spreads. The model, augmented by jumps to help explain short‐term spreads, is estimated on firm‐level data and identifies significant asset variance risk premia. An important feature of the model is the significant time variation in risk premia induced by the uncertainty about asset risk. Various extensions are considered, among them optimal leverage and endogenous default.

Rule-based credit risk assessment model using multi-objective evolutionary algorithms

Abstract: Credit risk assessment is considered as one of the vital topics in financial institutions. The existing credit risk evaluation methods are based on black box models or transparent models. The black box models cannot adequately reveal information hidden in the data and the credit risk evaluation remains difficult. In addition, there exist relatively few transparent models that take into consideration interpretability and comprehensibility. To address this problem, we aim to build a reliable credit risk evaluation model which generates a set of classification rules. In fact, we consider the credit risk evaluation as a search-based optimization problem where the goal is to minimize the complexity of the generated solution, to maximize the accuracy, and also to maximize weight which represents rules importance. We conducted a comparative study of four multi-objective evolutionary algorithms in terms of their performance. The obtained results confirm the efficiency of the SMOPSO Algorithm regarding generating classification rules for credit risk assessment. The proposed credit risk evaluation model revealed an attractive trade-off between accuracy and comprehensibility.

Detecting Financial Collapse and Ballooning Sovereign Risk

Abstract: This paper proposes a new model for capturing discontinuities in the underlying financial environment that can lead to abrupt falls, but not necessarily sustained monotonic falls, in asset prices This notion of price dynamics is consistent with existing understanding of market crashes, which allows for a mix of market responses that are not universally negative The model may be interpreted as a martingale composed with a randomized drift process that is designed to capture various asymmetric drivers of market sentiment In particular, the model is capable of generating realistic patterns of price meltdowns and bond yield inflations that constitute major market reversals while not necessarily being always monotonic in form The recursive and moving window methods developed in Phillips, Shi and Yu (2015a,b; PSY), which were designed to detect exuberance in financial and economic data, are shown to have detective capacity for such meltdowns and expansions This characteristic of the PSY tests has been noted in earlier empirical studies by the present authors and other researchers but no analytic reasoning has yet been given to explain why methods intended to capture the expansionary phase of a bubble may also detect abrupt and broadly sustained collapses The model and asymptotic theory developed in the present paper together explain this property of the PSY procedures The methods are applied to analyse S&P 500 stock prices and sovereign risk in European Union countries over 2001–16 using government bond yields and credit default swap (CDS) premia A pseudo real‐time empirical analysis of these data shows the effectiveness of the monitoring strategy in capturing key events and turning points in market risk assessment

Default prediction in P2P lending from high-dimensional data based on machine learning

Abstract: In recent years, a new Internet-based unsecured credit model, peer-to-peer (P2P) lending, is flourishing and has become a successful complement to the traditional credit business. However, credit risk remains inevitable. A key challenge is creating a default prediction model that can effectively and accurately predict the default probability of each loan for a P2P lending platform. Due to the characteristics of P2P lending credit data, such as high dimension and class imbalance, conventional statistical models and machine learning algorithms cannot effectively and accurately predict default probability. To address this issue, a decision tree model-based heterogeneous ensemble default prediction model is proposed in this paper for accurate prediction of customer default in P2P lending. Gradient boosting decision trees (GBDT), extreme gradient boosting (XGBoost) and light gradient boosting machine (LightGBM) are employed as individual classifiers to create a heterogeneous ensemble learning-based default prediction model. Learning model-based feature ranking is applied to P2P lending credit data, and individual classifiers undergo hyperparameter optimization. Finally, comparison with benchmark models shows that the prediction model can achieve desirable prediction results and thus effectively solve the challenge of predictions based on high-dimensional and imbalanced data.