The validity of combined risk models can be difficult to estimate without primary data. In this study, we introduce a novel algorithm designed specifically for this purpose. The algorithm requires just four parameters: the separation powers of two scorecards; the correlation between the two scorecards; and the default rate of the target population. By assuming an underlying multivariate normal structure, the algorithm returns either the estimated Gini coefficient or the area under the receiver operating characteristic curve of the combined scorecard. We examine the efficacy of this model through a series of Monte Carlo simulations, and test it using empirical data from four independent samples of consumer loans from financial institutions using a psychometric scorecard together with a traditional credit bureau scorecard. Overall, our model offers a practical tool for researchers and practitioners, and it illustrates the trade-offs in validity between the collinearity and discriminatory power of combined predictive models.
In modern society the pervasive use of credit cards has led to a substantial increase in fraudulent activities, resulting in considerable financial losses for both cardholders and issuing banks. Despite the efforts of academic and industry researchers to develop various algorithms for fraud detection, the multidimensional attributes of transaction data lack both a comprehensive exploration and applications. To bridge this gap this study introduces a model based on a stacked temporospatial graph attention residual network (stacked TS-GARN), specifically tailored for credit card fraud detection. The model first employs a temporal-spatial Node2Vec with attention (TS-N2VA) method to generate embedding vectors that capture the structural information of transaction nodes in both temporal and spatial dimensions. Subsequently, it integrates these temporal and spatial embedding vectors with their corresponding attribute features and further aggregates the features of neighboring transaction nodes through a multihead attention mechanism to produce a comprehensive feature vector. This vector is then used as input for subsequent classification tasks. For fraud detection, a two-layer residual network logistic stacking ensemble (TLSE) is used. Finally, experiments are conducted on two publicly available credit card data sets to evaluate the model's performance across various dimensions. The experimental results demonstrate that the stacked TS-GARN model effectively extracts latent features and achieves more accurate recognition outcomes than the other tested models.
Small businesses are shunned by the traditional banking system. In this context, microentrepreneurs are constantly developing tools to ensure the financing of their activities. Local financial engineering has led to the creation of the Adogbe` savings product in Benin. This paper provides an overview of and identifies the risks of the "Adogbe`" savings product, using a theoretical approach. By including the possibility of analyzing trade credit via informal savings products, the description of the Adogbe` product in this paper extends the reverse trade credit theory and hyperbolic discounting and regular installment theory. The paper analyzes the Adogbe` savings product as both a liability (for a fixed regular deposits collector, usually a supplier) and an asset (for a saver, usually a customer). The theoretical contribution of this research uses predictions obtained via Frank and Goyal's 2007 trade-off theory to advise microenterprises on how to achieve a balance between the benefits and costs or risks of using the Adogbe` savings product as a financing method. Adogbe` is both an effective strategy for sustainability and a risk factor for microenterprises. A new theory called "Adogbe` club hyperbolic discounting and regular installments" is formulated.
In this study we propose a formula-based approach for determining the optimal liquidity horizon used in scaling the base expected shortfall under the Basel Committee's market risk capital requirements. Specifically, the "half-life" formula introduced by Almgren and Chriss in 2000 is used to compute optimal liquidity horizon values and construct an optimal expected shortfall measure. Unlike the regulatory approach, which scales expected shortfall across aggregated groups of risk factors, as laid out in the Basel Committee's 2019 "Minimum capital requirements for market risk", our method enables scaling at the level of individual securities, thereby improving estimation accuracy. To evaluate performance, we compare optimal expected shortfall with the regulatory expected shortfall using returns data from the Standard & Poor's 500 and Dow Jones Industrial Average indexes. Employing the regression-based expected shortfall backtesting technique of Bayer and Dimitriadis, we find that optimal expected shortfall produces unbiased estimates, while regulatory expected shortfall systematically overstates the true magnitude of expected shortfall.
In supply chain management and trade credit, buyers of goods or services are often granted a delayed payment goal, and the sellers of the respective goods or services are thus exposed to credit risk. Factoring is a financing decision by which sellers can eliminate this risk from their balance sheet. This service is typically offered on a whole turnover basis; however, most small and medium-sized enterprises prefer financing on a single debt level, and hence adverse selection emerges as an additional source of risk. By employing a game-theoretical min-max approach including a Levy-frailty ansatz for the multivariate default model, we develop a hybrid linear pricing model for the factor, consisting of a static component and a dynamic component. As part of the derivation, we analyze a two-player Stackelberg game and develop a portfolio optimization problem for the supplier, based on a stochastic gradient descent algorithm, which results in an upper bound for the price. Combining this with a cost-covering condition based on the factor provides the proposed price interval. Finally, we embed into the model the simplest data-rich modeling approach for the dynamic part of our analysis.
This paper introduces a novel approach for allocating a bank's risk capital across individual portfolios and transactions. This allocation is pivotal for various capital applications, including risk-adjusted return on capital and lending decisions. Our proposed method, the hierarchy allocation method (HAM), overcomes several limitations inherent in traditional approaches such as the Euler principle. Notably, HAM is versatile enough to be applied to a broad spectrum of risk capital types, including both coherent measures, such as expected shortfall, and noncoherent measures, such as value-at-risk. A unique feature of our approach is its flexibility in accounting for diversification benefits, as it does not assume that combining portfolios will necessarily reduce risk. To provide a comprehensive understanding of risk capital, we also present a general mathematical definition. Empirical validation is conducted through data experiments that underscore the method's practical advantages and its capacity to handle different levels of correlation and diversification effects.
We propose a novel method of including macroeconomic variables in exposure at default models, which satisfies all expectations connected to International Financial Reporting Standard 9 requirements. In addition, it is intuitive and transparently transforms the situation in the credit environment into expected loss values. We propose a decomposition approach that separates the contract-based variables from the macroeconomic indicators. Using various estimation methods, we build a set of models that combine idiosyncratic information gathered at the exposure level with systematic indicators collected quarterly. We test our predictions on out-of-time data, which includes the Covid-19 pandemic period, and find that our decomposition outperforms benchmarks in terms of selected forecast quality metrics. The proposed solution allows risk managers to adjust capital levels, making their financial institution more competitive in the market.
The United States has become a global hot spot for payment (credit and debit) card fraud, with a notable portion of fraud incidents in the European Union (EU) linking back to the United States, particularly within the e-commerce ecosystem. In this paper we discuss the emergence of EMV (Europay, Mastercard and Visa) technology and its adoption in EMV payment cards, and the effectiveness of EMV cards in managing credit card fraud in the United States and EU. We cover the challenges and barriers faced during the implementation of EMV in the United States, as well as the lessons learned from Europe's earlier adoption. We also provide insights into the potential strategies and best practices to combat payment card fraud in the United States and EU. Ultimately, this study aims to contribute to the ongoing efforts to enhance the security of payment card transactions, protect consumers and maintain the integrity of the global financial ecosystem.
In 2016 the Bank of Russia developed two ordinances setting forth a procedure that uses a limited sample of loans to determine whether or not the level of loss provision for a portfolio of uniform loans is sufficient and whether the bank's capital is adequate. The procedure for assessing the adequacy of reserves, as a rule, involves considering only a part of the loan portfolio and extrapolating the reserves calculated in this way to the entire portfolio. Moreover, the procedure for determining the minimum sample size of loans assumes there is no default correlation. The contribution of our paper is the application of well-known, though often ignored, properties of the Bernoulli distribution of the total number of correlated events to a novel problem: an extrapolation of the capital provision that does not take into account the possible existence of a default correlation. As a result, we prove that the presence of a default correlation requires a larger minimum sample size of loans than when it is absent. More specifically, we justify how the minimum sample size of loans depends upon the absolute and relative differences in default rates (provision rates, rate of regulatory noncompliance) of two samples, the required significance levels and statistical power.
The assessment of the public interest rationale in bank resolution remains a contentious issue within the regulatory framework of the European Union (EU). The Bank Recovery and Resolution Directive (BRRD) was introduced to provide a structured approach to handling failing banks while minimizing systemic risk and protecting taxpayers. However, inconsistencies in the application of public interest assessments (PIAs) have led to divergent resolution practices across member states. This study analyzes 39 cases of banks classified as failing or likely to fail between 2015 and 2022, examining the key determinants influencing PIA decisions. Of these, the Single Resolution Board assessed public interest in 8 cases, while national resolution authorities handled 31 cases. This study empirically analyzes the key determinants of PIAs in EU bank resolution, focusing on systemic risk, bank size and bank localness. Using Firth's logistic regression, the analysis reveals that local banks are disproportionately subject to resolution, suggesting that national authorities may prioritize regional economic stability over strict systemic risk considerations. The findings highlight a regulatory bias that challenges the consistency and predictability of resolution decisions. This study contributes to the ongoing debate on banking resolution by identifying key determinants of PIA and proposing policy recommendations to enhance the transparency and uniformity of resolution practices across the EU.
This paper examines the impact of environmental, social and governance (ESG) disclosure on investment efficiency, leveraging the implementation of Directive 2014/ 95/EU as a quasi-natural experiment to assess changes in disclosure quality. It finds a significant and robust reduction in underinvestment among US firms with substantial operations in the European Union - and thereby subject to the Directive - compared with US-centric firms unaffected by the directive. The former experienced an increase in debt financing post-directive, though no significant changes are observed in the equity capital raised. The improvement in investment efficiency is most pronounced in firms with initially low ESG disclosure levels, those facing financial constraints and those with more entrenched managers. These findings imply that nonfinancial disclosure mandates, akin to financial reporting requirements, can alleviate capital rationing issues for underinvesting firms, particularly in debt markets. The study underscores the potential role of ESG disclosure in improving financial outcomes and informing policy on nonfinancial reporting standards.
Fintech lending to consumers has grown rapidly since the 2007-9 Great Recession. This study applies machine learning (ML) methods to loan-level data from the largest fintech lender of personal loans, to assess the extent to which these methods can produce more accurate out-of-sample default predictions relative to standard regression models, as argued by fintech lending's advocates. To explain loan outcomes, this analysis accounts for the economic conditions faced by a borrower after origination, which are typically absent from other ML studies of default. For the given data, the ML methods indeed improve prediction accuracy, but more so over horizons within a year. Having more data up to but not beyond a certain quantity enhances the relative predictive accuracy of the ML methods, likely because there has been data or model drift over time, so that more complex models can suffer more out-of-sample misses. Prediction accuracy rises, but only marginally, with additional standard credit variables beyond the core set, suggesting that unconventional data needs to be sufficiently informative as a whole to help consumers with little or no credit history.Finally, in this data, we find little statistically significant evidence that ML methods yield unequal benefits across subgroups of borrowers defined by their risk attributes, income or where they live.
The metaverse is a rapidly evolving concept centered on creating an ecosystem of interconnected virtual platforms, enabling its users to engage in real-world-like activities through digital avatars. In recent years, the metaverse's ability to create immersive experiences on these parallel virtual platforms has been increasingly appreciated by its users. However, creating and maintaining such metaverse platforms has entailed an extensive reliance on myriad unconventional and evolving technologies, which include hardware, such as highly sophisticated augmented reality or virtual reality headsets and accessories; software supported by extensive data to create and run the virtual platforms; and digital ledger technology to record transactions or execute smart contracts. Despite this, the metaverse has not attracted significant regulatory attention thus far, perhaps owing to its having few implications for financial stability given its current modest scale. This paper evaluates the vulnerabilities in the metaverse stemming from technological inadequacies, unregulated data collection, the dominance of Big Tech and the risks of fraud and manipulation. It also draws attention to emerging issues, such as market conduct, integrity and investor protection in the metaverse, which may necessitate supervisory attention.
