In this paper, we study the option-implied systemic risk measures through a model-free approach. Employing this approach, we estimate systemic risk measures using ETF options on the S&P 500 index and its nine industry sectors. The empirical study finds that the option-implied systemic risk measures enable earlier risk monitoring and more accurate responses to market shifts. Backtesting results demonstrate that option-implied systemic risk measures exhibit superior predictive accuracy for risk compared to their counterparts based on historical returns (hereafter referred to as underlying-asset-based measures). The gaps between underlying-asset-based and risk-neutral systemic risk measures during the COVID-19-related market downturn significantly predict the excess returns of the S&P 500 index. Besides, we apply these risk measures for portfolio optimization, showing that portfolios optimized with option-implied information achieve superior performance compared to those based on historical data.

This study investigates the safe-haven characteristics of gold and 12 major cryptocurrencies in the context of macroeconomic uncertainty. The analysis categorizes uncertainty into two primary dimensions: systemic financial stress and global policy instability. We expand set of Heterogeneous Autoregressive (HAR) models, which incorporate asymmetric effects, quantile-on-quantile methodologies, and Lasso-based variable selection to capture nonlinearities and high-dimensional dynamics. The results show that gold volatility is significantly influenced by uncertainty indicators, while most cryptocurrencies exhibit no significant response to uncertainty. The asymmetric analysis suggests that upward uncertainty shocks disproportionately affect volatility in both asset classes. Gold emerges as a superior safe haven during high volatility, while certain cryptocurrencies show resilience during moderate volatility. The Lasso-based selection identifies that gold volatility is influenced by external uncertainty, while cryptocurrency volatility is driven solely by internal factors. Overall, cryptocurrencies are found to serve better as diversifiers than traditional safe havens, offering important insights for risk management and portfolio adjustments in times of global uncertainty.

The network null model is a statistical framework for making inferences about complex network systems based on partially observed attributes, identifying statistically significant patterns within networks. This study extends the null model for weighted bipartite networks by introducing diversification heterogeneity constraints (measured via the Herfindahl-Hirschman index) within the maximum entropy framework. Compared to traditional degree-or strength-constrained null models, our approach captures edge weight dispersion heterogeneity and could find broad application in ecology, economics, and management science. To address the associated numerical challenges, we develop a low-rank initialization scheme and an alternating optimization algorithm. Simulation study confirms the distinctive advantages of the new model and the effectiveness of the numerical solution. In addition, our empirical analysis shows that China's financial institutions' overlapping equity investment can be reproduced by the proposed model.

This paper examines whether the implied volatility surface contains incremental predictive return information beyond conventional option-implied characteristics. We construct standardized delta--maturity grids of implied volatilities and apply machine learning methods to predict future stock excess returns. Over the 2011--2024 out-of-sample period, H--L portfolios formed on the resulting surface-based signals generate sizable Sharpe ratios and significant alphas relative to standard factor benchmarks, outperforming most hand-crafted option variables. Further tests show that nonlinear models exploit broader surface patterns and retain significant alphas after controlling for option-implied variables and factor-mimicking portfolios associated with economic channels. The evidence suggests that the implied volatility surface contains incremental predictive information that is not fully captured by existing low-dimensional option characteristics.

This study examines cojump dynamics through network modeling, analyzing both positive and negative cojumps using 5-min high-frequency data from 194 stocks within the CIS 300 index. We apply eigenvector centrality to identify key stocks within these networks and utilize community detection method to classify clusters of stocks exhibiting similar cojump patterns. The findings indicate that negative cojump network demonstrates stronger inter-stock linkages, identifies the most influential stocks in finance and real estate sectors, and exhibits greater community intensity compared to the positive cojump network. Portfolios constructed using negative cojump rankings achieve higher Sharpe ratios than those based on positive cojump networks, and the integrating of negative cojump community information further improves return predictions. Overall, these insights underscore the vital role of negative cojump dynamics in optimizing investment strategies and strengthening risk management.

This study introduces a method for computing daily systemic risk measures using high-frequency data, specifically realized conditional value-at-risk (RCoVaR) and realized marginal expected shortfall (RMES). RCoVaR and RMES are empirical estimators derived from scaling high-frequency returns, offering benefits such as model-independence and adaptability to diverse datasets. To mitigate market microstructure noise (MMN) inherent in high-frequency data, we employ overlapping and subsampling approaches in the estimation of RCoVaR and RMES. Empirical analysis focuses on systemic risk within the foreign exchange market. The results indicate that noise-treated RCoVaR and RMES serve as effective alternatives for daily systemic risk estimation. These techniques also enhance out-of-sample predictive accuracy when employed as predictors within systemic risk forecasting frameworks.

In this study, we construct a novel measure of daily trade-based stock manipulation, stock manipulation intensity (SMI), by training a machine learning model on a comprehensive sample of 159 prosecuted manipulation cases pursued by the China Securities Regulatory Commission (CSRC) between 2010 and 2023. The SMI effectively captures complex, nonlinear dynamics and interactions among high-frequency trading variables, providing a powerful tool for regulatory authorities to detect manipulation. We further show that the SMI is distinct from herding and momentum effects, capturing micro-structural information these proxies miss and dominating both as a predictor of manipulation. In the full-sample empirical analysis, both the SMI and direct punishment cases reveal three consistent patterns. First, manipulation is associated with significantly elevated trading activity and price volatility. Second, we observe a classic ``pump-and-dump'' pattern, where manipulated stocks exhibit substantial short-term abnormal returns followed by significant reversals. Third, compared with state-owned enterprises (SOEs), non-state-owned enterprises (non-SOEs) face significantly greater manipulation.

In this paper, we propose a price staleness factor model that accounts for pervasive market friction across assets and incorporates relevant covariates. Using large-panel high-frequency data, we derive the maximum likelihood estimators of the regression coefficients, the nonstationary factors, and their loading parameters. These estimators recover the time-varying price staleness probabilities. We develop asymptotic theory in which both the dimension d and the sampling frequency n tend to infinity. Using a local principal component analysis (LPCA) approach, we find that the efficient price co-volatilities (systematic and idiosyncratic) are biased downward due to the presence of staleness. We provide bias-corrected estimators for both the spot and integrated systematic and idiosyncratic co-volatilities, and prove that these estimators are robust to data staleness. Interestingly, besides their dependence on the dimensionality d, the integrated plug-in estimates converge at a rate of n^-1/2 without requiring correcting term, whereas the local PCA estimates converge at a slower rate of n^-1/4. This validates the aggregation efficiency achieved through nonlinear, nonstationary factor analysis via maximum likelihood estimation. Numerical experiments justify our theoretical findings. Empirically, we demonstrate that the staleness factor provides unique explanatory power for cross-sectional risk premia, and that the staleness correction reduces out-of-sample portfolio risk.

PurposeThis paper aims to identify the macro variables that affect China's ETS market through a mixed-frequency sampling data with a variable selection model.Design/methodology/approachThis paper focuses on the Hubei, Guangdong and Shenzhen ETS in China. It integrates exogenous factors, in aspect of economic, financial, energy and environment, to identify key drivers of ETS market volatility. First, this study applies the GARCH-MIDAS model to process mixed-frequency data. Second, this paper employs the Lasso method to select the most predictive factors and enhance volatility forecasting. Finally, this paper evaluates the effectiveness of the conclusions through model parameter estimation, out-of-sample prediction, robustness test and economic value evaluation.FindingsFirst, China's ETS market volatility is primarily driven by the energy sector, with limited influence from policy and environmental factors. Second, ETS market volatility varies across regions. The power sector strongly influences the Hubei ETS market, whereas the Guangdong and Shenzhen ETS markets are more affected by the energy market. Third, out-of-sample analysis and robustness tests statistically indicate that the GARCH-MIDAS-Adaptive-Lasso model enhances forecasting accuracy.Originality/valueFirst, this paper integrates multidimensional factors into the model. Second, this paper combines the adaptive Lasso method with the GARCH-MIDAS model to analyze the volatility of China's ETS market. This method addresses both multicollinearity and variable selection challenges in mixed-frequency data. Third, this paper offers valuable insights for other developing countries seeking to establish or enhance ETS systems.

This paper introduces a high-dimensional binary variate model that accommodates nonstationary covariates and factors, and studies their asymptotic theory. This framework encompasses scenarios where single indices are nonstationary or cointegrated. For nonstationary single indices, the maximum likelihood estimator (MLE) of the coefficients has dual convergence rates and is collectively consistent under the condition T^1/2/N→0, as both the cross-sectional dimension N and the time horizon T approach infinity. The MLE of all nonstationary factors is consistent when T^δ/N→0, where δ depends on the link function. The limiting distributions of the factors depend on time t, governed by the convergence of the Hessian matrix to zero. In the case of cointegrated single indices, the MLEs of both factors and coefficients converge at a higher rate of min(√(N),√(T)). A distinct feature compared to nonstationary single indices is that the dual rate of convergence of the coefficients increases from (T^1/4,T^3/4) to (T^1/2,T). Moreover, the limiting distributions of the factors do not depend on t in the cointegrated case. Monte Carlo simulations verify the accuracy of the estimates. In an empirical application, we analyze jump arrivals in financial markets using this model, extract jump arrival factors, and demonstrate their efficacy in large-cross-section asset pricing.

Since the implementation of the Basel III Accord, expected shortfall (ES) has gained increasing attention from regulators as a complement to value-at-risk (VaR). The problem of elicitability for ES makes jointly modeling VaR and ES a popular method to study ES. In this article, we develop model averaging for joint VaR and ES regression models that selects the two weight vectors by minimizing a jackknife criterion. We show the large sample properties of the estimators under potential model misspecification with increasing dimension of parameters and the asymptotic optimality of the selected weights in the sense of minimizing the out-of-sample excess final prediction error. Simulation studies and three empirical analyses reveal good finite sample performance.

We conduct a comparative analysis of quantitative models for assessing risk contagion and systemic risk within the Chinese financial market, focusing on four key methodologies: vector autoregression-forecast error variance decomposition (VAR-FEVD), quantile vector autoregression-forecast error variance decomposition (QVAR-FEVD), linear conditional value-at-risk (CoVaR) and tail-event driven network (TENET). Our analysis underscores the significance of network construction methods in accurately depicting the spillover effects among financial institutions. The research delves into the performance of financial networks by comparing "physical" networks, evaluating predefined networks within the dynamic network quantile regression model, and identifying systemically important financial institutions across various network configurations. In particular, the TENET model emerges as particularly adept, outperforming the other models in capturing both mean and tail risk spillovers. This paper not only deepens the understanding of systemic risk in China but also provides valuable recommendations for policy makers to design effective regulatory frameworks to mitigate potential crises.

Download This Paper Open PDF in Browser Add Paper to My Library Share: Permalink Using these links will ensure access to this page indefinitely Copy URL Copy DOI

This paper focuses on simulation-based approaches for estimating systemic risk measures. In particular, we provide the asymptotic forms of the relative errors for widely used systemic risk measures includ-ing conditional value-at-risk (CoVaR), coexpected shortfall (CoES) and marginal expected shortfall (MES). Based on asymptotic expansions, a general framework is provided for the simulation of systemic risk measures. The numerical results show that the proposed simulation framework works well, and it is more user-friendly, easier to expand and less time-consuming than simulation approaches using the re -sampling method and importance sampling. (c) 2023 Elsevier B.V. All rights reserved.

The single-index model (SIM) reveals the intricate relationship between the response variable and covariates, allowing for the consideration of potential heterogeneity and nonlinearity in the data, while also offering interpretability and flexibility. This paper concentrates on estimating the parameters and the unknown link function for the quantile SIM with high-dimensional covariates. We introduce a novel iterative algorithm that strikes a balance between estimation accuracy, computational efficiency, and adaptability to diverse datasets. The initial values for iteration and iteration termination conditions are also discussed. The finite sample performance is illustrated through a simulation study, demonstrating the advantages of our method in terms of accuracy and speed.

This paper explores the tail risk network in the Chinese green-related stock market, by estimating the Copula-MIDAS-LASSO model. The model integrates characteristic factors and improves the model’s ability to capture the tail risk spillover. Additionally, we utilize the MTS network and the threshold network to picture the network in the market. The findings reveal that securities consistently hold central positions in the risk contagion network, with GFS, CUB, and PAI being three key sources of risk contagion. Finally, the subsample analysis demonstrates that the green finance policy and financial crises contribute to increased risk dependence within the market.

This paper proposes a novel model for the valuation of variance swaps that incorporates a multi-factor stochastic spot variance and a multi-factor stochastic long-term variance, while allowing mean reversion in the asset price and a co-jump structure in the model. We propose a general analytical approach for pricing discretely monitored variance swaps via a moment-based method and confirm its accuracy and efficiency using Monte Carlo simulations. Our empirical results indicate that the model with a three-factor spot variance and a one-factor long-term variance significantly outperforms other nested models. The incorporation of multiple factors in our model is essential not only for fitting market data, but also for reconciling the term structure of variance swaps.