In highly class-imbalanced financial distress prediction, standard classifiers often fail as they assume balanced class distributions and equal misclassification costs. We propose a heterogeneous voting ensemble method that addresses extreme class imbalance by synergizing sampling, cost-sensitive learning, and ensemble techniques. First, coordinating sampling and cost-sensitive learning reduces model bias in identifying distressed firms. Sampling rebalances class distributions, which reduces representation bias and provides an effective data foundation for cost-sensitive learning. Cost-sensitive learning assigns higher costs to distressed firms, adjusting the optimization objective to minimize total misclassification costs. Second, by aggregating diverse base classifiers, the heterogeneous voting ensemble reduces potential additional variance introduced by sampling and cost-sensitive learning, thereby improving generalization. Using data from China, Poland, and the United States, we show that the proposed method significantly outperforms benchmark models in detecting distressed firms. Moreover, the improved predictions translate into reduced misclassification costs, particularly when the cost difference between distressed and healthy firms is greater. Ablation experiments validate the effectiveness of each component, where sampling contributes most to detection accuracy for distressed firms, followed by cost-sensitive learning, while the ensemble ensures model stability. These results highlight that combining sampling, cost-sensitive learning, and ensemble techniques is crucial for stakeholders (e.g., investors, creditors, and regulators) to detect rare events such as financial distress.

The high degree of interconnectedness and interdependence within the global financial system implies that cross-border spillovers from monetary policy propagate through complex transmission channels. This study uses a global vector autoregression (GVAR) model, using monthly data from 17 G20 economies over the period 2008-2023, to systematically investigate how US quantitative easing (QE) monetary policy, proxied by two key spread measures, affects global financial markets. The first proxy, the term spread, is defined as the yield on 30-year US Treasury securities minus that on 10-year Treasury securities; the second, the mortgage spread, is defined as the 30-year fixed mortgage rate minus the 10-year Treasury yield. A negative shock to either spread (that is, a narrowing) captures the interest rate compression typically induced by QE. The empirical results reveal that a 1% decline in the 30-year Treasury yield is associated with an increase of up to 5% in 10-year government bond yields in advanced economies, while the most pronounced effects in emerging markets occur in foreign exchange markets, where exchange rate responses reach as high as 23%. Moreover, the mortgage spread exerts a stronger influence across regional financial markets than the term spread, consistently generating more significant spillover effects. The Chicago Board Options Exchange volatility index (VIX) and monetary policy uncertainty (MPU) emerge as the two dominant transmission channels. Our analysis highlights substantial heterogeneity in cross-country responses, particularly in foreign exchange markets. Furthermore, the study examines the moderating roles of financial leverage and commodity exposure, showing that these factors significantly shape the magnitude and persistence of spillover effects.

This study proposes a novel generalized dynamic factor tail-restricted integrated regression function (GDF-IRF) network to investigate the idiosyncratic tail risk spillover in Chinese commodity futures markets. The core advance of this model lies in its design as a directed tail-conditional dependence network, which is built on factor-filtered idiosyncratic components and aggregated over short-, medium-, and long-term horizons. Using data from 1,604 trading days spanning March 2018 to November 2024, we analyze idiosyncratic tail risk transmission patterns across different frequency domains and crisis periods (e.g., the US-China trade friction, the COVID-19 pandemic, and the Russia-Ukraine war). Our results show that: (1) The contagion in the idiosyncratic tail risk network is significantly higher than that in the tail risk network. Furthermore, the former is stronger in the upper tail than in the lower tail, whereas the latter follows the opposite trend. (2) Coal-related and soybean futures are the main idiosyncratic risk transmitters in the lower and upper tails, while oil-related and soybean meal futures act as idiosyncratic risk receivers in the lower and upper tails. (3) The regression results indicate that both commodity characteristics and macroeconomic factors drive futures’ contagiousness, but their effects are asymmetric. Investors and policymakers could use our findings as early warning tools to identify influential risk spreaders during crisis periods.

This paper investigates the optimal portfolio of industry stocks with China's green bond (CBGB) at different time scales, through comparing minimum connectedness portfolio (MCoP1) and modified minimum connectedness portfolio (MCoP2) under the condition of minimizing extreme risk connectedness, as well as minimum VaR portfolio. First, we examine the ability of CBGB to provide diversification benefits for industry stocks at different scales, which employs extreme risk connectedness. Then, the portfolio performing optimally at each scale is explored by comparing specific performances of various portfolio methods. The empirical results show that: (1) CBGB is weakly related to industry stocks, whether in the time or frequency domain, and is the net receiver. (2) Most industry stocks play different roles (transmitter or receiver) in different scales, with time-varying characteristics. Therefore, industry stocks benefit highly from the diversification effects of CBGB, and their portfolio should possess frequency and time-varying characteristics. (3) Based on the comparison results of weight allocations, cumulative returns, and Sharpe ratios, the short-term optimal performance is achieved by MCoP2, while in medium- to long-term, MCoP1 outperforms MCoP2 and minimum VaR portfolio. This implies that considering extreme risk connectedness in the construction of portfolios can effectively enhance returns, but in the long term, market integration should also be taken into account.

With network topology measures, we can model the global and individual properties of international extreme sovereign risk spillovers and understand how shocks propagate. Hence, using dynamic connectedness based on a TVP-VAR model, we construct daily extreme sovereign risk spillover networks based on defined extreme risk series among the G20. Our purpose is to creatively explore the network structures and describe international connectedness. We find that system- and country-level measures are all more sensitive to global systemic events, such as the COVID-19 pandemic. Country-level analysis shows that emerging countries such as Mexico, South Africa and European countries such as Spain emit and receive larger risk spillovers. Using the Logit and threshold regression, we creatively explore whether these system- and country-level measures can explain the probability of countries’ extreme sovereign risk outbreaks. The results show that system-level measures such as total risk spillover strength and country-level measures all play positive roles. Specifically, the greater the total risk spillover strength, the more central countries’ position and the greater the probability of countries’ extreme sovereign risk outbreak. Most importantly, their roles are the largest when the total risk spillover strength is at the middle level.

This study innovatively investigates the threshold effects in the impacts of chronic and acute climate risks on bank stability in China by using the dynamic panel threshold model. The results indicate that chronic risks in high-risk intervals and acute risks in low-risk intervals exert significant adverse effects on bank stability. Rural commercial banks, larger, lower-capitalized and lower-liquid banks, as well as banks located in less developed cities are more vulnerable. Moreover, the increase in climate policy uncertainty helps to mitigate the impacts of climate risks and maintain financial stability.

Investor sentiment connectedness (TCI), which depicts the aggregate effect of investor sentiment contagion, is documented as a potential risk contagion channel and plays an essential role in the evolution of systemic risk. This study applies the TVP-VAR model to construct TCI and examines whether TCI can predict banking systemic risk. Empirical results show that an increase in TCI strongly and stably predicts increasing banking systemic risk over the next month. Additionally, both in-sample and out-of-sample results indicate that TCI provides incremental information for banking systemic risk prediction.

Theoretically, international EPU spillovers would amplify negative impacts of particular shocks on countries’ financial markets and macro level, which may drive sovereign risk and its spillovers. Hence, we explore EPU spillovers among G20, and investigate spatial spillovers of sovereign risk from the perspective of EPU spillovers and explore its influential factors during various periods. First, we construct the EPU spillover network based on EPU connectedness obtained by the TVP-VAR model. Network analysis results show crisis has exacerbated EPU spillovers. And developed countries are net exporters of EPU spilloversSecond, based on the database among G20 over the period 2008Q3-2019Q4, we innovatively construct the EPU spillover matrix as spatial weight matrix and apply dynamic spatial SAR model. There are significant spatial spillovers containing contemporaneous and time-lagged interactions in sovereign risk during the European debt crisis period and only contemporaneous interactions during the financial crisis period. Last, we analyze contemporaneous and diffusion effects based on unit changes and country specific variation in the macro-economic variable. We find the much higher magnitude and variation in contemporaneous effects based on country specific variation.

This paper employs a spatial econometrics method to study the international spillovers of economic policy uncertainty (EPU), and contributes to the literature by exploring the transmission channels. We comprehensively consider bilateral trade, financial investment, and information channels (government liability, trade imbalance, fiscal imbalance, business cycle). Based on a large sample of 21 countries from 2001Q1 to 2021Q4, our analysis indicates that the channels proposed are all significantly effective, and bilateral trade contributes the most. We also investigate the dominant channels during five crisis and non-crisis periods: in the global financial crisis, financial investment and government liability similarity are of utmost importance; in European debt crisis, bilateral trade and information channels of government liability and business cycle are the main channels; before these crises, the dominant channel is financial investment, which turned to be bilateral trade in the post-crises period; in COVID-19 pandemic, financial investment is the most important.

基于我国各商业银行的银行间资产和银行间负债总体数据,利用最大熵法间接推断得到银行间借贷关联网络.在此基础上,利用DebtRank算法研究了单一冲击和共同冲击情形下我国商业银行损失困境的传染过程,进而衡量银行的系统重要性和系统脆弱性及其影响因素.研究结果表明,在特定情形下,银行损失困境给系统内其他银行造成的权益损失反而比银行违约造成的损失更大.我国商业银行没有位于"高脆弱性/强重要性"区域,表明我国银行体系较为安全.在影响因素方面,平均资产回报率对银行的系统重要性具有显著的负向影响,贷款拨备率、一级资本充足率和同业拆借率均对银行的系统重要性具有显著的正向影响;一级资本充足率和资产规模对银行的系统脆弱性均具有显著负向影响,同业拆借率对银行的系统脆弱性具有显著的正向影响.研究结果不仅有助于各银行了解自身处境,而且为金融监管提供依据.

With the globalisation of the economy and financial markets, cross-market risk spillovers have become increasingly prominent, affecting global financial stability. Using the Diebold-Yilmaz Connectedness Index (DYCI) method, we construct interacting networks to explore these risk spillovers across four multi-country markets: sovereign credit default swap (SCDS), stock, foreign exchange, and commodity markets. We find: First, at the system-level, multiple inter-layer connections between the SCDS and other markets demonstrate the high level of cross-market risk spillovers. They are sensitive to major economic and financial events, such as the COVID-19 pandemic and Russia-Ukraine conflict. SCDS has the closest spillovers with stocks. The block model analysis shows that all relationships between various blocks contain inter-layer connections, and the SCDSs of emerging countries and stocks of developed countries are the sources of risk spillover in the entire system. Second, at the country-level, SCDS' cross-market spillover strength changes by country and time. Five cross-node centrality analysis shows that the SCDSs of emerging countries are the risk spillover engines between the SCDS and other markets. The spillovers of a country's SCDS to gold (oil) are more sensitive to the Fed rate hike in 2016 (the European debt crisis and Russia-Ukraine conflict). Third, to understand the driving factors, we perform time series and panel regressions for cross-market risk spillovers and cross-node centrality, respectively. The influences of economic fundamentals and market sentiment are asymmetric in high- and low-risk spillovers. The greater the total economic linkages, such as trade and capital flows, the greater are the countries' cross-market risk spillovers.

Considering the dual risks of extreme downside liquidity and extreme negative sentiment, we introduce the concept of the joint lower-tail risk of liquidity and investor sentiment (LISR) and construct measures to study the issue of lower-tail risk premiums in the Chinese stock market. Our findings provide convincing evidence that the premiums for LISR measures are significant regardless of the sentiment at the market or firm level. Downside liquidity risk and extreme negative sentiment cannot explain the LISR premiums separately, which means that an extreme downside change in liquidity and extreme negative sentiment have a joint effect on future stock returns. In addition, LISR premiums are robust to various portfolio double sorts, hold for various asset pricing factor models and remain significant when controlling for an extensive list of firm characteristics. Our conclusions have obvious value for improving and enriching the theoretical research on investor sentiment and liquidity risk premiums, and they provide a valuable reference for investors aiming to construct portfolios matching their own risk preferences, and for regulators supervising the market.

This paper uses the VMD-Copula-& UDelta;CoVaR model to examine the dynamic characteristics of extreme risk spillovers between the oil market and BRICS stock markets during COVID-19 from a multiscale perspective. Using daily data from January 1, 2019 to December 31, 2021, the empirical results show that: (1) Bidirectional extreme risk spillovers exist between oil and BRICS stock markets for the original return series, short-term, medium-term, and long-term components, with BRICS stock markets generally exporting higher risk to the oil market than they receive. (2) The magnitude of risk spillovers at all time scales presents an inverted V-shaped pattern during COVID-19, increasing with the worsening of the pandemic and the oil price war, then decreasing after various countermeasures. (3) Following the COVID-19 vaccine rollout, the long-term spillovers between oil and stock markets in countries such as Brazil, China, and South Africa have returned to or even fallen below pre-pandemic levels. Conversely, the magnitude of short-term and medium-term risk spillovers generally remains higher than pre-COVID-19 levels. (4) The short- and medium-term components are the primary contributors to overall risk spillovers, whereas the long-term component contributes the least. (5) Oil and the Russian stock market exhibit the strongest risk spillovers at all time scales, while oil and the Chinese stock market have the weakest risk spillovers. Overall, our findings suggest that investors and policymakers should consider the impact of the COVID-19 pandemic and various countermeasures on the multiscale spillovers between oil and BRICS stock markets to optimize decision-making and maintain financial market stability.

Purpose This study is the first that aims to investigate international transmission channels of sovereign risk among G20 and explore its influential factors by applying the multidimensional SAR model. Design/methodology/approach Multiple spatial weight matrices can capture the contiguity of spatial units from various dimensions, which could be exploited to improve the precision of inference as well as prediction accuracy. To the best of the authors’ knowledge, this is the first study to investigate international transmission channels of sovereign risk among G20 and explore its influential factors by applying the multidimensional SAR model. Findings With network structure analysis, this study finds that they contain different information content from the perspective of graphical display, node strength and correlation. Developed and emerging countries all play major roles in trade connection, while only developed countries play major roles in financial linkage. Second, by applying the multidimensional SAR model, only the spatial autocorrelation coefficients for trade and financial linkages are significant during the full sample period, which is in sharp contrast to published studies using the SAR model with a single matrix. Third, the spillover channels that play major roles in various periods are different. Only trade channel plays a role during crisis periods and it is the most important. Fourth, the spatial correlation among countries greatly amplifies the shock’s impacts on one market. And spatial effect for developed countries is larger than those for emerging countries, while the mean spatial effect of a unit shock in the USA on emerging countries is slightly greater than that on developed countries. Originality/value Multiple spatial weight matrices can capture the contiguity of spatial units from various dimensions, which could be exploited to improve the precision of inference as well as prediction accuracy. To the best of the authors’ knowledge, this is the first study to investigate international transmission channels of sovereign risk among G20 and explore its influential factors by applying the multidimensional SAR model.

This article investigates for the first time the role of idiosyncratic tail risk in the cross-sectional pricing of bond returns. We use the idiosyncratic return of the bond to measure the idiosyncratic tail risk based on the extreme value theory. The results show that bonds in the highest idiosyncratic tail risk quintile generate 3.5% more annual return compared to bonds in the lowest idiosyncratic tail risk quintile. In addition, we found that idiosyncratic tail risk is cross-sectionally positive correlated with bond expected returns in Chinese bond market, even when the downside risk, bond rating, liquidity, size, maturity, bond market beta, short-term reversals, and coupon rate are controlled. The positive correlation is in line with the traditional risk-return tradeoff theory. Because of their aversion to extreme losses, investors are willing to accept the low returns from low idiosyncratic tail risk bonds.

Based on the asymmetric distribution and extreme distribution of liquidity, we construct the third-moment and the fourth-moment of liquidity to measure the downside liquidity risk and extreme liquidity risk respectively, and study the liquidity premium through high-moment measures. The results show that the high moments of liquidity contain more pricing information that cannot be captured by the low-moment measures. Downside liquidity risk has a long-term and continuous premium effect, and extreme liquidity risk has a short-term premium effect. Furthermore, idiosyncratic risk and market risk are the main driving factors of downside liquidity risk premium and extreme liquidity risk premium respectively.

This article applies quantile regression approaches to investigate the asymmetric effects of economic policy uncertainty (EPU) on G7 stock returns by dividing EPU changes into EPU increases and EPU decreases. We observe that EPU increases have greater impacts on G7 stock returns than EPU decreases, which confirms that asymmetric effects do exist. Furthermore, our results show that EPU changes have asymmetric impacts on Japan, Canada, the UK, and the US stock returns in the bearish market but not in the bullish market. In addition, there are no asymmetric effects of EPU changes on Germany stock returns.

Systemic risk emphasizes the impact on the real economy and is popularly measured by a network interconnectedness approach. We test, for the first time, whether the volatility connectedness of financial institutions is a significant predictor of Chinese macroeconomy. The connectedness is derived from volatility spillover networks and is measured by total connectedness introduced in Diebold and Yilmaz (2014), which reflects the effects of risk transmission and systemic risk in the financial system. Both in-sample and out-of-sample analyses show that an increase in total connectedness among financial institutions stably and strongly forecasts a slowdown in China's economic activity over the next three to twelve months, when controlling for many factors. Furthermore, including the total connectedness into the regression models improves the macroeconomy forecasts accuracy. Our results are robust to alternative measures of total connectedness.

文章基于制造业整体特征,选取财务指标和非财务指标构建制造企业信用风险预警指标体系;在传统BP神经网络模型的基础上对输入指标运用因子分析进行降维优化处理,通过蝙蝠算法优化传统BP神经网络的随机权值问题,构建制造业上市公司信用风险预警的FA-BA-BP模型;以60家制造业上市公司为样本,进行实证研究.通过对数据的降维处理,提炼出盈利能力、偿债能力、运营能力、自身规模、创新能力和经营规范六个信用风险预警因子.研究表明:FA-BA-BP神经网络模型迭代次数更少、均方误差和平均绝对误差均为最小,且预测准确率达到95％,具有良好的预警作用.

We apply novel spatial econometric techniques to investigate spillovers in sovereign risk for 41 advanced and emerging economies during 2004-2019. We find that sovereign risk spillover channels that play major roles in various periods are different. Real linkages (trade, financial, and geography) and information channel both play major roles in spillover effects during the full sample period. During the financial crisis period, only business connections (trade and financial) have an effect, while only the geographical distance channel did not have an effect during the European debt crisis period. We also assess the relative importance of the direct and indirect effects of macroeconomic variables. We observe that the long-term effects are larger than the short-term effects. Ultimately, our analysis emphasizes the importance of considering multiple channels when analyzing sovereign risk spillovers.