|
on Risk Management |
| By: | Haoyuan Wang; Chen Liu; Minh-Ngoc Tran; Chao Wang |
| Abstract: | This paper introduces a novel multivariate volatility modeling framework, named Long Short-Term Memory enhanced BEKK (LSTM-BEKK), that integrates deep learning into multivariate GARCH processes. By combining the flexibility of recurrent neural networks with the econometric structure of BEKK models, our approach is designed to better capture nonlinear, dynamic, and high-dimensional dependence structures in financial return data. The proposed model addresses key limitations of traditional multivariate GARCH-based methods, particularly in capturing persistent volatility clustering and asymmetric co-movement across assets. Leveraging the data-driven nature of LSTMs, the framework adapts effectively to time-varying market conditions, offering improved robustness and forecasting performance. Empirical results across multiple equity markets confirm that the LSTM-BEKK model achieves superior performance in terms of out-of-sample portfolio risk forecast, while maintaining the interpretability from the BEKK models. These findings highlight the potential of hybrid econometric-deep learning models in advancing financial risk management and multivariate volatility forecasting. |
| Date: | 2025–06 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2506.02796 |
| By: | Arishi Orra; Aryan Bhambu; Himanshu Choudhary; Manoj Thakur; Selvaraju Natarajan |
| Abstract: | Portfolio optimization requires dynamic allocation of funds by balancing the risk and return tradeoff under dynamic market conditions. With the recent advancements in AI, Deep Reinforcement Learning (DRL) has gained prominence in providing adaptive and scalable strategies for portfolio optimization. However, the success of these strategies depends not only on their ability to adapt to market dynamics but also on the careful pre-selection of assets that influence overall portfolio performance. Incorporating the investor's preference in pre-selecting assets for a portfolio is essential in refining their investment strategies. This study proposes a volatility-guided DRL-based portfolio optimization framework that dynamically constructs portfolios based on investors' risk profiles. The Generalized Autoregressive Conditional Heteroscedasticity (GARCH) model is utilized for volatility forecasting of stocks and categorizes them based on their volatility as aggressive, moderate, and conservative. The DRL agent is then employed to learn an optimal investment policy by interacting with the historical market data. The efficacy of the proposed methodology is established using stocks from the Dow $30$ index. The proposed investor-specific DRL-based portfolios outperformed the baseline strategies by generating consistent risk-adjusted returns. |
| Date: | 2025–04 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2505.03760 |
| By: | Muhammed Golec; Maha AlabdulJalil |
| Abstract: | Large Language Models (LLM), which have developed in recent years, enable credit risk assessment through the analysis of financial texts such as analyst reports and corporate disclosures. This paper presents the first systematic review and taxonomy focusing on LLMbased approaches in credit risk estimation. We determined the basic model architectures by selecting 60 relevant papers published between 2020-2025 with the PRISMA research strategy. And we examined the data used for scenarios such as credit default prediction and risk analysis. Since the main focus of the paper is interpretability, we classify concepts such as explainability mechanisms, chain of thought prompts and natural language justifications for LLM-based credit models. The taxonomy organizes the literature under four main headings: model architectures, data types, explainability mechanisms and application areas. Based on this analysis, we highlight the main future trends and research gaps for LLM-based credit scoring systems. This paper aims to be a reference paper for artificial intelligence and financial researchers. |
| Date: | 2025–06 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2506.04290 |
| By: | Joseph Lu; Randall R Rojas; Fiona C. Yeung; Patrick D. Convery |
| Abstract: | We design a portfolio construction framework and implement an active investment strategy utilizing momentum and trend-following signals across multiple asset classes and asset class risk factors. We quantify the performance of this strategy to demonstrate its ability to create excess returns above industry standard benchmarks, as well as manage volatility and drawdown risks over a 22+ year period. |
| Date: | 2025–06 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2506.09330 |
| By: | Hans Buehler; Blanka Horvath; Yannick Limmer; Thorsten Schmidt |
| Abstract: | This paper addresses the challenge of model uncertainty in quantitative finance, where decisions in portfolio allocation, derivative pricing, and risk management rely on estimating stochastic models from limited data. In practice, the unavailability of the true probability measure forces reliance on an empirical approximation, and even small misestimations can lead to significant deviations in decision quality. Building on the framework of Klibanoff et al. (2005), we enhance the conventional objective - whether this is expected utility in an investing context or a hedging metric - by superimposing an outer "uncertainty measure", motivated by traditional monetary risk measures, on the space of models. In scenarios where a natural model distribution is lacking or Bayesian methods are impractical, we propose an ad hoc subsampling strategy, analogous to bootstrapping in statistical finance and related to mini-batch sampling in deep learning, to approximate model uncertainty. To address the quadratic memory demands of naive implementations, we also present an adapted stochastic gradient descent algorithm that enables efficient parallelization. Through analytical, simulated, and empirical studies - including multi-period, real data and high-dimensional examples - we demonstrate that uncertainty measures outperform traditional mixture of measures strategies and our model-agnostic subsampling-based approach not only enhances robustness against model risk but also achieves performance comparable to more elaborate Bayesian methods. |
| Date: | 2025–06 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2506.07299 |
| By: | Jose Portela; Eduardo S. Schwartz; Jaime Aparicio Garcia |
| Abstract: | Legal claims are increasingly being considered as an alternative asset class, however, there appears to be a lack of a standard methodology for valuing litigation risk. This paper proposes a dynamic real options framework for the valuation of legal claims, explicitly incorporating the uncertainty and sequential nature of litigation processes. We develop a continuous-time stochastic model that accounts for the main procedural milestones and uncertainties, enabling the simulation of diverse litigation trajectories to estimate the net present value of a claim. The model permits the decision-maker to optimally continue or abandon the litigation at various stages, thereby capturing the embedded option value and enhancing claim valuation. This approach offers a novel risk management and valuation tool for a range of stakeholders, including investors, third-party funders, claimants, defendants, legal practitioners, auditors, and insurers. We demonstrate the practical relevance of the methodology by applying it to an actual international investment arbitration case. |
| JEL: | G01 G11 K0 |
| Date: | 2025–05 |
| URL: | https://d.repec.org/n?u=RePEc:nbr:nberwo:33790 |
| By: | YANG, ZHANG (Department of Finance and Business Economics, Faculty of Business Administration / Asia-Pacific Academy of Economics and Management, University of Macau); JIANXIONG LIN (QIFU Technology, China); YIHE QIAN (Department of Finance and Business Economics, Faculty of Business Administration, University of Macau); LIANJIE SHU (Faculty of Business Administration , University of Macau) |
| Abstract: | MachiAs a key enabler of poverty alleviation and equitable growth, financial inclusion aims to expand access to credit and financial services for underserved individuals and small businesses. However, the elevated default risk and data scarcity in inclusive lending present major challenges to traditional credit assessment tools. This study evaluates whether machine learning (ML) techniques can improve default prediction for small-business loans, thereby enhancing the effectiveness and fairness of credit allocation. Using proprietary loan-level data from a city commercial bank in China, we compare eight classification models—Logistic Regression, Linear Discriminant Analysis (LDA), K-Nearest Neighbors (KNN), Support Vector Machine (SVM), Decision Tree, Random Forest, XGBoost, and LightGBM—under three sampling strategies to address class imbalance. Our findings reveal that undersampling significantly enhances model performance, and tree-based ML models, particularly XGBoost and Decision Tree, outperform traditional classifiers. Feature importance and misclassification analyses suggest that documentation completeness, demographic traits, and credit utilization are critical predictors of default. By combining robust empirical validation with model interpretability, this study contributes to the growing literature at the intersection of machine learning, credit risk, and financial development. Our findings offer actionable insights for policymakers, financial institutions, and data scientists working to build fairer and more effective credit systems in emerging markets. |
| Keywords: | machine learning, financial inclusion, small business, China, credit risk assessment |
| JEL: | G21 G32 C53 O16 |
| Date: | 2025–06 |
| URL: | https://d.repec.org/n?u=RePEc:boa:wpaper:202532 |
| By: | Roman Goncharenko (KU Leuven - Department of Accountancy, Finance and Insurance (AFI); Central Bank of Ireland); Mikhail Mamonov (TBS Business School); Steven Ongena (University of Zurich - Department Finance; Swiss Finance Institute; KU Leuven; NTNU Business School; Centre for Economic Policy Research (CEPR)); Svetlana Popova (The Central Bank of Russian Federation); Natalia Turdyeva (Bank of Russia) |
| Abstract: | How do firms respond to sudden and forcible closures of their lenders? Using unique credit register data from a setting where two-thirds of banks were closed within a decade, we find that neither bad nor good firms delay repayments or switch lenders before closures. Afterward, bad firms lose subsidized credit and experience sharp declines in employment, borrowing, and sales, while good firms improve performance. This divergence stems from banks’ prior underpricing of bad firms’ credit risk. Ultimately, good firms match with new solid banks, while bad firms gravitate toward not-yet-detected weak banks---especially where boards overlap or markets are unconcentrated. |
| Keywords: | Firms, Bank clean-up policies, Regulatory forbearance, Credit risk underpricing, Common board membership, Real effects |
| JEL: | G21 G28 G32 L25 |
| Date: | 2025–05 |
| URL: | https://d.repec.org/n?u=RePEc:chf:rpseri:rp2552 |
| By: | Pia Stoczek (Paderborn University); Alexander Liss (KU Leuven); Boaz Noiman (The Hebrew University of Jerusalem) |
| Abstract: | We examine risk-taking by lending syndicates as a response to central banks’ corporate quantitative easing (QE) targeting non-financial firms, specifically within the European Central Bank’s Corporate Sector Purchase Programme (CSPP). This setting allows us to investigate how syndicates adjust to decreased credit demand from CSPP-eligible borrowers in environments characterized by higher risk and lower returns. Our analysis reveals that these syndicates engage in “controlled” risk-taking by directing capital towards first-time and non-relationship borrowers, especially in the leveraged loan sector, while implementing mechanisms to manage increased risk. Our study explores controlled risk-taking across four dimensions. Firstly, we observe adjustments in loan contracting terms, such as stricter collateral requirements and cross-default clauses, coupled with reductions in loan sizes and maturities. Secondly, our findings indicate that syndicate size and the intensity of relationships within syndicates increase. Thirdly, we highlight the influence of the borrower country’s debt enforcement regime on lending decisions. Lastly, we report no significant changes in loan spreads. These results suggest that following corporate QE, syndicates actively utilize risk mitigation mechanisms, demonstrating a cautious approach to managing elevated risks rather than excessive risk-taking. |
| Keywords: | Loan contracting, Relationship lending, Unconventional monetary policy, Quantitative easing |
| JEL: | E52 E60 G12 G21 G28 G30 |
| Date: | 2025–05 |
| URL: | https://d.repec.org/n?u=RePEc:pdn:dispap:142 |
| By: | Michal Franta; Jan Vlcek |
| Abstract: | Inflation at Risk provides a coherent description of the risks associated with an inflation outlook. This paper explores the practical applicability of this approach in central banks. The method is applied to Czech inflation to highlight issues related to short data sample. A set of quantile regressions with a non-crossing quantiles constraint is estimated using monthly data from the year 2000 onwards, and the model's in-sample fit and out-of-sample forecasting performance are then assessed. Furthermore, we discuss the Inflation at Risk estimates in the context of several historical events and demonstrate how the approach can inform monetary policy. The estimation results suggest the presence of nonlinearities in the Czech inflation process, which are related to supply-side pressures. In addition, it appears that regime changes have occurred recently. |
| Keywords: | Inflation dynamics, inflation risk, quantile regressions |
| JEL: | E31 E37 E52 |
| Date: | 2025–05 |
| URL: | https://d.repec.org/n?u=RePEc:cnb:wpaper:2025/8 |
| By: | Paige Ehresmann; Juan M. Morelli; Jessie Jiaxu Wang |
| Abstract: | In this note, we introduce a factor asset pricing model to analyze risk-adjusted returns on bank stocks. Given their high-frequency availability, bank stock returns offer a valuable lens into the risk exposures and dynamics of the banking sector. |
| Date: | 2025–06–06 |
| URL: | https://d.repec.org/n?u=RePEc:fip:fedgfn:2025-06-06-3 |
| By: | Matt Brigida; Kathleen Maceyka |
| Abstract: | In this analysis we determine factors driving the cross-sectional variation in uninsured deposits during the interest rate raising cycle of 2022 to 2023. The goal of our analysis is to determine whether banks proactively managed deposit run risk prior to the hiking cycle which produced the 2023 Regional Banking Crisis. We find evidence that interest rate forward, futures, and swap use affected the change in a bank uninsured deposits over the period. Interest rate option use, however, has no effect on the change in uninsured deposits. Similarly, bank equity levels were uncorrelated with uninsured deposit changes. We conclude we find no evidence of banks managing run risk via their balance sheet prior to the 2023 Regional Banking Crisis. |
| Date: | 2025–06 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2506.03344 |
| By: | Sergio Correia; Stephan Luck; Emil Verner |
| Abstract: | Why do banks fail? We create a panel covering most commercial banks from 1863 through 2024 to study the history of failing banks in the United States. Failing banks are characterized by rising asset losses, deteriorating solvency, and an increasing reliance on expensive noncore funding. These commonalities imply that bank failures are highly predictable using simple accounting metrics from publicly available financial statements. Failures with runs were common before deposit insurance, but these failures are strongly related to weak fundamentals, casting doubt on the importance of non-fundamental runs. Furthermore, low recovery rates on failed banks' assets suggest that most failed banks were fundamentally insolvent, barring strong assumptions about the value destruction of receiverships. Altogether, our evidence suggests that the primary cause of bank failures and banking crises is almost always and everywhere a deterioration of bank fundamentals. |
| Date: | 2025–06 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2506.06082 |
| By: | Metz-Peeters, Maike; Patragst, Jil-Laurel |
| Abstract: | This study presents an exploratory analysis of the key factors contributing to fatal and severe crashes on German motorways. We employ Poisson and Negative Binomial regression models, combined with Lasso regularization and stability selection, to explore model specifications incorporating potentially many interaction terms and polynomials. Utilizing an extensive data set including rich geo-spatial characteristics for 500-meter segments covering large parts of the German motorway network, key variables influencing crash frequency are uncovered. To obtain correct standard errors post variable selection, we split the data into separate samples for model selection and parameter estimation. Our results indicate that the inclusion of a limited number of higher-order terms significantly improves the regression formulation. Robustness checks confirm the stability of these findings. The results offer clearer insights into the key crash determinants and are more computationally feasible than simulation-based methods commonly used in accident research. |
| Abstract: | Diese Studie präsentiert eine explorative Analyse von Schlüsselfaktoren, die zu tödlichen und schweren Unfällen auf deutschen Autobahnen beitragen. Wir verwenden Poisson- und Negativ-Binomial-Regressionen, kombiniert mit Lasso-Regularization und Stability Selection, um Modellspezifikationen mit potenziell zahlreichen Interaktionstermen und Polynomen zu untersuchen. Basierend auf einem umfangreichen Datensatz, der raumbezogene Merkmale für 500-Meter-Abschnitte des deutschen Autobahnnetzes enthält, werden zentrale Variablen identifiziert, die die Unfallhäufigkeit beeinflussen. Um korrekte Standardfehler nach der Variablenselektion zu gewährleisten, teilen wir die Daten in separate Stichproben für Modellauswahl und Parameterschätzung. Unsere Ergebnisse zeigen, dass die Einbeziehung einer begrenzten Anzahl Terme höherer Ordnung die Regressionsformulierung signifikant verbessert. Robustheitstests bestätigen die Stabilität dieser Erkenntnisse. Die Resultate bieten Einblick in die zentralen Unfalldeterminanten und sind rechnerisch effizienter als simulationsbasierte Methoden, die in der Unfallforschung üblicherweise eingesetzt werden. |
| Keywords: | Road safety, crash frequency, lasso regression, machine learning, stability selection |
| JEL: | C52 H10 R41 |
| Date: | 2025 |
| URL: | https://d.repec.org/n?u=RePEc:zbw:rwirep:319076 |