Consumer credit-risk models via machine-learning algorithms

We apply machine-learning techniques to construct nonlinear nonparametric forecasting models of consumer credit risk. By combining customer transactions and credit bureau data from January 2005 to April 2009 for a sample of a major commercial bank’s customers, we are able to construct out-of-sample forecasts that significantly improve the classification rates of credit-card-holder delinquencies and defaults, with linear regression R²’s of forecasted/realized delinquencies of 85%. Using conservative assumptions for the costs and benefits of cutting credit lines based on machine-learning forecasts, we estimate the cost savings to range from 6% to 25% of total losses. Moreover, the time-series patterns of estimated delinquency rates from this model over the course of the recent financial crisis suggest that aggregated consumer credit-risk analytics may have important applications in forecasting systemic risk.     

Modelling failure rates with machine-learning models: Evidence from a panel of UK firms

In this study, we investigate the ability of machine-learning techniques to predict firm failures and we compare them against alternatives. Using data on business and financial risks of UK firms over 1994–2019, we document that machine-learning models are systematically more accurate than a discrete hazard benchmark. We conclude that the random forest model outperforms other models in failure prediction. In addition, we show that the improved predictive power of the random forest model relative to its counterparts persists when we consider extreme economic events as well as firm and industry heterogeneity. Finally, we find that financial factors affect failure probabilities.     

Stock investment strategy combining earnings power index and machine learning

We propose an intermediate-term stock investment strategy based on fundamental analysis and machine learning. The approach uses predictors from the Earnings Power Index (EPI) as input variables derived from cross-sectional and time-series data from a company’s financial statements. The analytical methods of machine learning allow us to validate the link between financial factors and excess returns directly. We then select stocks for which returns are likely to increase at the time of the next disclosed financial statement. To verify the proposed approach’s usefulness, we use company data listed publicly on the Korean stock market from 2013 to 2019. We examine the profitability of trading strategy based on ten machine-learning techniques by forming long, short, and hedge portfolios with three different measures. As a result, most portfolios, including EPI-related variables, present positive returns regardless of the period. Especially, the neural network of the two layers with sigmoid function presents the best performance for the period of 3 months and 6 months, respectively. Our results show that incorporating machine learning is useful for mid-term stock investment. Further research into the possible convergence of financial statement analysis and machine-learning techniques is warranted.     

Capturing downside risk in financial markets: the case of the Asian Crisis

Using data on Asian equity markets, we observe that during periods of financial turmoil, deviations from the mean-variance framework become more severe, resulting in periods with additional downside risk to investors. Current risk management techniques failing to take this additional downside risk into account will underestimate the true Value-at-Risk with greater severity during periods of financial turnoil. We provide a conditional approach to the Value-at-Risk methodology, known as conditional VaR-x, which to capture the time variation of non-normalities allows for additional tail fatness in the distribution of expected returns. These conditional VaR-x estimates are then compared to those based on the RiskMetrics™ methodology from J.P. Morgan, where we find that the model provides improved forecasts of the Value-at-Risk. We are therefore able to show that our conditional VaR-x estimates are better able to capture the nature of downside risk, particularly crucial in times of financial crises.     

Data-driven optimization of peer-to-peer lending portfolios based on the expected value framework

In recent years, peer-to-peer (P2P) lending has been gaining popularity amongst borrowers and individual investors. This can mainly be attributed to the easy and quick access to loans and the higher possible returns. However, the risk involved in these investments is considerable, and for most investors, being nonprofessionals, this increases the complexity and the importance of investment decisions. In this study, we focus on generating optimal investment decisions to lenders for selecting loans. We treat the loan selection process in P2P lending as a portfolio optimization problem, with the aim being to select a set of loans that provide a required return while minimizing risk. In the process, we use internal rate of return as the measure of return. As the starting point of the model, we use machine-learning algorithms to predict the default probabilities and calculate expected values for the loans based on historical data. Afterwards, we calculate the distance between loans using (i) default probabilities and, as a novel step, (ii) expected value. In the calculations, we utilize kernel functions to obtain similarity weights of loans as the input of the optimization models. Two optimization models are tested and compared on data from the popular P2P platform Lending Club. The results show that using the expected-value framework yields higher return.     

基于余弦相似度(CS)和最小生成树(MST)的基金市场复杂网络研究

以中国基金市场中123家基金公司持有的投资组合为样本,综合运用余弦相似度(CS)和最小生成树(MST)方法,考量基金市场复杂网络。结果显示:各家基金公司持有股票组合的相似程度比持有债券组合的相似程度更高,表明他们持有的债券组合较之股票组合更加多元化,基金公司持有的股票相对集中于市值大、成长性高的公司。同时,全部资产投资组合、股票投资组合和债券投资组合等三类基金MST网络的节点度均服从幂律分布,表明大多数基金公司以少数强影响力基金公司为中心聚集起来,彼此之间具有较强的业务关联。此种网络结构特征可能导致市场风险向基金聚集团体集中,其抵御系统性风险的能力偏弱,也不利于满足投资者的理财多元化需求。     

CFO role and CFO compensation: An empirical analysis of their implications

Given concerns over CFO pay, especially incentives, and considering the tension between a CFO’s fiduciary responsibility and being a key member of the firm’s executive team, we examine the determinants and effects of CFO compensation amount, incentive intensity, and proximity to CEO compensation in a sample of European companies (FTE 500, 2005–2009). First, we focus on the CFO role as a determinant of CFO compensation. Like prior work, we proxy for CFO roles by using hand-collected public data on education and past professional experience, but we supplement these proxies with proprietary data to more directly capture the firm-specific nature of the CFO job in term of its similarity with that of the CEO. We thus argue how CFOs can have varied roles characterized by different levels of financial expertise and CEO-likeness, and document that it is this latter aspect that is associated with CFO compensation. Second, we study the effects of CFO compensation design on outcomes in the CFO’s realm related to financial reporting. We find that CFO financial expertise is positively associated with financial reporting quality, while a CFO’s pay long-term incentive intensity and a CFO’s incentive compensation proximity with the CEO are negatively associated with financial reporting quality. Overall, then, our results suggest that CFOs get rewarded for their CEO-likeness, and particularly for their being similar to the CEO in terms of tasks and decision making authority. But it is their financial expertise that is positively related to financial reporting quality. At the same time, using compensation that is more incentive intensive and more similar to that of the CEO appears to be potentially detrimental to the quality of financial reporting. These results are relevant for boards involved in selecting highly expert CFOs, and their compensation committees charged with defining subsequently effective incentive compensation plans for those CFOs.     

Identifying systemically important financial institutions in Turkey

This paper examines the systemic risk of financial firms in Turkey. Using Component Expected Shortfall, we provide estimates of systemic risk in Turkey using daily data from 2005 to 2018 and a comprehensive data set encompassing 54 financial firms. Empirical results show that the preponderance of systemic risk in the sample in Turkey is due to large commercial banks. Top ten systemically important financial institutions dominate systemic risk measures in Turkey and account for more than 90 % of total risk over the sample. Consequently, the risk in the Turkish financial system is concentrated in specific financial institutions and makes close monitoring of the top firms essential. Historical incidence of systemic risk in the sample shows elevated levels of systemic risk correspond to well-known external events. Finally, a bivariate VAR model shows that systemic risk is correlated with measures of global financial risks and has significant negative effects on the real economy particularly on industrial production. This is important from a financial stability point of view in that close monitoring of the systemic risk is important in maintaining a healthy financial system and a well- functioning market economy.     

Detecting stock market manipulation via machine learning: Evidence from China Securities Regulatory Commission punishment cases

In this paper, we apply machine-learning techniques to construct detecting models of stock market manipulation. By combining manually collected China Securities Regulatory Commission punishment cases from 2014 to 2016 with financial information of listed companies, we construct a training set and a test set to compare the detecting ability of support vector machine (SVM) and logistic model. Considering imbalanced data, we further incorporate Borderline Synthetic Minority Oversampling Technique (Borderline SMOTE) to oversample minority class and then find that Borderline SMOTE–SVM performs better than SVM and benchmark model in detecting manipulation. To enhance detecting performance of the models, we innovatively introduce market sentiment indicators which are extracted from analyst rating reports, financial news, and Guba comments into our indicators set. The results indicate that the new indicators generate significant marginal increment to the model accuracy.     

Value-at-risk and the cross section of emerging market hedge fund returns

This paper examines the cross-sectional relationship between downside risk (Value at Risk) and expected returns in a sample of 1370 emerging market hedge funds (EMHF). We find that downside risk significantly drives expected returns for these funds, particularly before the global financial crisis, commanding an annual risk premium of over 12%. While EMHF differ from their advanced market counterparts in risk/return patterns, we show that the global financial crisis of 2008 has caused a structural shift in that pattern. Finally, we show that the risk premium associated with downside risk is predictable by the global financial cycle, even after we control for emerging market systematic risk factors.     

Regionality revisited: An examination of the direction of spread of currency crises

What determines the direction of spread of currency crises? We examine data on waves of currency crises in 1992, 1994, 1997, and 1998 to evaluate several hypotheses on the determinants of contagion. We simultaneously consider trade competition, financial links, and institutional similarity to the “ground zero” country as potential drivers of contagion. To overcome data limitations and account for model uncertainty, we utilize Bayesian methodologies hitherto unused in the empirical literature on contagion. In particular, we use the Bayesian averaging of binary models that allows us to take into account the uncertainty regarding the appropriate set of regressors.We find that institutional similarity to the ground zero country plays an important role in determining the direction of contagion in all the emerging market currency crises in our dataset. We thus provide persuasive evidence in favour of the “wake-up call” hypothesis for financial contagion. Trade and financial links may also play a role in determining the direction of contagion, but their importance varies amongst the crisis periods.     

Systemic Risk Contributions

We adopt a systemic risk indicator measured by the price of insurance against systemic financial distress and assess individual banks’ marginal contributions to the systemic risk. The methodology is applied using publicly available data to the 19 bank holding companies covered by the U.S. Supervisory Capital Assessment Program (SCAP), with the systemic risk indicator peaking around $1.1 trillion in March 2009. Our systemic risk contribution measure shows interesting similarity to and divergence from the SCAP loss estimates under stress test scenarios. In general, we find that a bank’s contribution to the systemic risk is roughly linear in its default probability but highly nonlinear with respect to institution size and asset correlation.     

Openness,financial structure,and bank risk: International evidence

Using panel data from a large cross-country sample covering 97 countries over the period 1996–2017, we combine 2SLS procedure with system GMM estimation to study the relationship between openness, financial structure and bank risk. The main contribution of the paper is that we identified a new channel, i.e. the financial structure channel, through which financial openness reduces bank risk. In particular, we find that as financial openness increases, a country's financial system tends to be more market-based, and a more market-based financial system is associated with higher bank market power, better information sharing and more revenue diversification, all of which contribute to the reduction in bank risk. We also find that the effect of inflow restrictions on bank risk is more pronounced than that of outflow restrictions. These results highlight the importance of an appropriate design of a country's opening-up strategy to match the evolution of its financial structure to increase bank stability.     

Predicting credit card delinquencies: An application of deep neural networks

The objective of this paper is twofold. First, it develops a prediction system to help the credit card issuer model the credit card delinquency risk. Second, it seeks to explore the potential of deep learning (also called a deep neural network), an emerging artificial intelligence technology, in the credit risk domain. With real-life credit card data linked to 711,397 credit card holders from a large bank in Brazil, this study develops a deep neural network to evaluate the risk of credit card delinquency based on the client's personal characteristics and the spending behaviours. Compared with machine-learning algorithms of logistic regression, naive Bayes, traditional artificial neural networks, and decision trees, deep neural networks have a better overall predictive performance with the highest F scores and area under the receiver operating characteristic curve. The successful application of deep learning implies that artificial intelligence has great potential to support and automate credit risk assessment for financial institutions and credit bureaus.     

Tail dependence and indicators of systemic risk for large US depositories

In this study, we investigate the extreme loss tail dependence between stock returns of large US depository institutions. We find that stock returns exhibit strong loss dependence even in their limiting joint extremes. Motivated by this result, we derive extremal dependence-based systemic risk indicators. The proposed systemic risk indicators reflect downturns in the US financial industry very well. We also develop a set of firm-level average extremal dependence measures. We show that these firm-level measures could have been used to identify the firms that were more vulnerable to the 2007–2008 financial crisis. Additionally, we explore the performance of selected systemic risk indicators in predicting the crisis performance of large US depository institutions and find that the average stock return correlations are also good predictors of crisis period returns. Finally, we identify factors predictive of extremal dependence for the US depository institutions in a panel regression setting. Strength of extremal dependence increases with asset size and similarity of financial fundamentals. On the other hand, strength of extremal dependence decreases with capitalization, liquidity, funding stability and asset quality. We believe the proposed indicators have the potential to inform the prudential supervision of systemic risk.     

The effects of board gender diversity on a firm's risk strategies

We study whether board gender diversity (BGD) affects corporate risk strategies. Specifically, we investigate the association between BGD and firms’ reputation risk and financial risk. Using S&P data from 1997 to 2013, we find that BGD is negatively associated with tax avoidance, suggesting firms with gender‐diverse boards are more cautious about potential reputation risks associated with aggressive tax strategies. However, we find that BGD is positively associated with firms’ financial risk. The combined findings illustrate that BGD aligns a firm's risk exposure closer to risk‐neutral shareholders’ preferences by reducing reputation risk exposure while enabling necessary financial risk exposure.     

Effects of Health and Longevity on Financial Risk Tolerance

We investigate the effects of health and life expectancy on tolerance of financial risk. Using a standard life-cycle model, we find that the effects of health and life expectancy on preferences over lifetime-income risk are theoretically ambiguous. However, risk tolerance is independent of health and life expectancy when utility takes one of the standard (harmonic absolute risk aversion) functional forms or when optimal consumption is constant over time. Our empirical results, using data from a stated-preference survey (n=2,795), suggest that financial risk tolerance is positively associated with both health and life expectancy; hence utility is not consistent with standard functional forms.     

Tail Risk Networks of Insurers Around the Globe: An Empirical Examination of Systemic Risk for G-SIIs vs Non-G-SIIs

In this article, we investigate systemic risk of 157 insurers around the globe. We construct tail risk networks among these insurers using a single-index model for quantile regressions with a variable selection technique. We develop a new network-based systemic risk indices, taking into account expected tail losses of insurers, direct and indirect contagion effects, and the time-varying strength of tail risk spillover. Our systemic risk indices successfully recognize global systemically important insurers (G-SIIs). We find that on average G-SIIs are more systemically relevant than non-G-SIIs, particularly during the recent U.S. financial crisis. We also find a small group of non-G-SIIs that are more important than G-SIIs. Our results have significant implications for systemic risk regulation.     

Risk spillovers between FinTech and traditional financial institutions: Evidence from the U.S.

In this paper, we propose a novel approach to examine the risk spillovers between FinTech firms and traditional financial institutions, during a time of fast technological advances. Based on the stock returns of U.S. financial and FinTech institutions, we estimate pairwise risk spillovers by using the Granger causality test across quantiles. We consider the whole distribution: the left tail (bearish case), the right tail (bullish case) and the center of the distribution and construct three types of spillover networks (downside-to-downside, upside-to-upside, and center-to-center) and obtain network-based spillover indicators. We find that linkages in the network are stronger in the bearish case when the risk of spillover is higher. FinTech institutions' risk spillover to financial institutions positively correlates with financial institutions' increase in systemic risk. These results have important policy implications, as they underscore the importance of enhancing the supervision and regulation of FinTech companies, to maintain financial stability.