This study develops a novel deep learning framework for bank distress prediction, designed to overcome the limitations of static network analysis and to enhance model interpretability. We propose a Spatio-Temporal Attention Network that uniquely captures the time-varying nature of systemic risk. Methodologically, it introduces two key innovations: (1) a dynamic interbank network whose connection weights are adjusted by the volatility of the Shanghai Interbank Offered Rate (SHIBOR), reflecting real-time market liquidity changes; and (2) a dual spatio-temporal attention mechanism that identifies critical time steps and pivotal contagion pathways leading to a distress event. Empirical results demonstrate that the model significantly outperforms traditional benchmarks across key metrics including accuracy and F1-score. Most critically, the architecture proves exceptionally effective at reducing Type II errors, substantially minimizing the failure to identify at-risk banks. The model also offers high interpretability, with attention weights visualizing intuitive risk evolution patterns. We conclude that incorporating dynamic, liquidity-adjusted networks is crucial for superior predictive performance in systemic risk modeling.

This study develops a novel deep learning framework for bank distress prediction, designed to overcome the limitations of static network analysis and to enhance model interpretability. We propose a Spatio-Temporal Attention Network that uniquely captures the time-varying nature of systemic risk. Methodologically, it introduces two key innovations: (1) a dynamic interbank network whose connection weights are adjusted by the volatility of the Shanghai Interbank Offered Rate (SHIBOR), reflecting real-time market liquidity changes; and (2) a dual spatio-temporal attention mechanism that identifies critical time steps and pivotal contagion pathways leading to a distress event. Empirical results demonstrate that the model significantly outperforms traditional benchmarks across key metrics including accuracy and F1-score. Most critically, the architecture proves exceptionally effective at reducing Type II errors, substantially minimizing the failure to identify at-risk banks. The model also offers high interpretability, with attention weights visualizing intuitive risk evolution patterns. We conclude that incorporating dynamic, liquidity-adjusted networks is crucial for superior predictive performance in systemic risk modeling.

Individual institutions marginal contributions to the systemic risk contain predictive power for its potential future exposure and provide early warning signals to regulators and the public. We use higher-order co-skewness and co-kurtosis to construct systemic risk contribution measures, which allow us to identify and characterize the co-movement driving the asymmetry and tail behavior of the joint distribution of asset returns. We illustrate the usefulness of higher-order moment augmented approach by using 4868 stocks living in the Chinese market from June 2002 to March 2022. The empirical results show that these higher-order moment measures convey useful information for systemic risk contribution measurement and portfolio selection, complementary to the information extracted from a standard principal components analysis.

This paper analyzes the amplification characteristics of risk propagation in multi-layer financial networks from the perspective of network topologies. The research finds that, even without feedback pathways, cross-layer links alone can lead to more severe risk contagion than in single-layer networks. More importantly, with increasing connectivity, the formation of cyclic structures in multi-layer networks will significantly exacerbate the systemic risk.

Climate change shocks pose a threat to the stability of the financial system. This study examines the influence of climate risks on systemic risk in the Chinese market by utilizing extreme risk spillover network. Moreover, we construct climate risk indices for physical risks (abnormal temperature), and transition risks (Climate Policy Uncertainty). We demonstrate a significant increase in systemic risk due to climate risks, which can be attributed, in part, to investor sentiment. Furthermore, institutional investors can mitigate the adverse impact of climate risks. Our findings suggest that policymakers and investors need to exercise greater vigilance in addressing climaterelated adverse effects.

This paper examines the impact of collateral shocks on corporate financialization using a sample of Chinese-listed firms from 2008 to 2021. We find a statistically and economically significant positive effect of collateral appreciation on financialization, consistent with profit-chasing motives, even after addressing endogeneity concerns. Additional tests reveal the effects are more pronounced among financially constrained, bank-dependent, and high-agency-cost firms. Financialization also elevates the risktaking and financial risks of firms. Overall, we provide novel evidence that collateral shocks stimulate corporate financialization, with implications for incentives, regulation, and systemic risk monitoring.

This study shows that investment decisions drive tail risks (i.e., systemic risk and stand-alone tail risk) of TBTF (Too-Big-to-Fail) banks, while financing decisions determine tail risks of non-TBTF banks. After the Dodd-Frank Act, undercapitalized non-TBTF banks continue to gamble for resurrection, and their stand-alone tail risk become more sensitive to funding availability and net-stable-funding-ratio than TBTF banks. We show that implementing a slimmed-down version of TBTF regulations on non-TBTF banks cannot efficiently contain the stand-alone risk of non-TBTF banks and cannot eliminate TBTF privilege. Moreover, non-TBTF banks together generate larger pressure of contagion on the real economy, and they herd more when making financing decisions after the Act. Our findings highlight the need for enhanced regulations on the liability-side of non-TBTF banks.

Based on the formation and evolution of systemic risk, this study proposes high and low volatility spillover networks and explores the characteristics of the evolution of systemic risk in Chinese financial market, and identifies the source of risk accumulation and risk outbreak, as well as the corresponding contagion mechanisms. Moreover, a new multiscale decomposition method (MVMD) is used to decompose high and low volatility into different time frequency components (short-term and long-term), and the corresponding network is constructed. Upon comparing topological characteristics on each layer from system and individual levels, our results reveal that high and low volatility spillover networks have different network characteristics and evolution behaviors. At the individual level, bond market is always the largest risk net-receivers in the high and low volatility networks, while the futures market and the currency market are respectively risk net-emitters in the high and low volatility networks. Additionally, compared with high volatility network, the low volatility network has greater predictive ability for financial risk. Finally, frequency analysis demonstrates that high-low volatility networks have different spillover intensity and network structure at different time frequencies. The above findings are beneficial for policy makers and investors to formulate appropriate strategies in different evolution of systemic risk and time frequency.

This paper argues that ffnancial lease, a dominant representation of shadow banking in China, plays a special role in improving the capital allocation efficiency. In a two-sector general equilibrium model with heterogeneous firm, information asymmetry and financial frictions, this paper shows that existence of finance lease market increases aggregate TFP by allowing low productivity SOE firms to lend out and allowing high productivity POE firms to leverage up. Due to the repossession advantage, financial leasing is a “good“ form of shadow banking that does not necessarily cause financial systemic risks.

China frequently suffers from weather related natural disasters and is a source of wide-spread systemic risk throughout large swaths of China. During these periods farmers crops are at risk and for a largely poor population few can afford the turmoil to livelihoods that goes along with drought. Throughout the developing world there is serious interest in index-based weather insurance for agriculture, and in China the China Insurance Regulatory Commission is investigating the insurability of weather related risk. Beyond that little formal research has appeared on either the demand, use or design of index insurance in China. This paper provides a preliminary assessment of farmers’ willingness to pay for drought insurance. Based on a survey of over 890 farm households in Shaanxi and Gansu provinces the results show that while there is significant demand, price may be an issue. Our results show that the majority of farm households would transition from a no-demand state to a demand state as prices fall. This suggests that in order to gain wide gain adoption there may be a need for governmental intervention.