Yingxiu Zhao

Accurate carbon price forecasting is essential for emission trading risk management and low-carbon investment decisions. In existing decomposition-prediction frameworks, secondary decomposition targets are typically selected based on statistical complexity rather than domain-informed causality, and standard Long Short-Term Memory (LSTM)-Transformer architectures discard the cell state that encodes long-term temporal memory. These limitations are particularly pronounced where energy-driven causal structures and regime-switching volatility coexist. This study proposes Causal State-driven Network (CaStNet), an intelligent forecasting framework with two core innovations. A Policy-Causality-guided Residual Secondary Decomposition (PCRSD) module replaces entropy-based criteria with Granger causality to select intrinsic mode functions (IMFs) exhibiting significant energy-carbon causal linkages for targeted variational mode decomposition (VMD). A Cell-State-Driven Dual-function Attention (CSDA) mechanism repurposes the LSTM cell state for simultaneously injecting long-term memory into the Transformer and employing the cell-state differential velocity as a volatility proxy to adaptively regulate Top-k attention sparsity. The Artificial Lemming Algorithm (ALA) globally co-optimizes decomposition dimensions and attention boundaries. A Shapley Additive exPlanations (SHAP)–Local Interpretable Model-agnostic Explanations (LIME) interpretability analysis reveals horizon-dependent driver transitions from short-term autoregressive momentum to long-term energy fundamentals, uncovering threshold nonlinearities in energy-carbon transmission channels. Validation on the Shanghai market (2013–2025) achieves point-forecast RMSE = 0.8326 and R² = 0.9777, outperforming all twelve benchmark models. Cross-market testing on the Hubei market yields R² = 0.9487, and expanding-window five-fold cross-validation on the Shanghai dataset yields mean R² = 0.9704, jointly confirming generalization robustness.

Utilizing panel data spanning from 2010 to 2021 across 30 Chinese provinces, this study examines the spatiotemporal dynamics of green finance and its correlation with high-quality economic development. Initially, the indicators for green finance and high-quality economic development were quantified by employing the entropy weight method. Secondly, we conducted a preliminary analysis of the spatiotemporal evolution patterns of green finance using the Mann-Kendall coefficient method and kernel density estimation, revealing an increasing trend in China’s green finance development level with regional disparities. Next, utilizing the Mann-Kendall coefficient method and spatial dynamic Markov model, we analyzed the spatiotemporal trends of green finance development and its coupling coordination with high-quality economic development across the 30 provinces in China. The research findings indicate a general upward trend in the degree of coordinated development between green finance and high-quality economic development from 2010 to 2021. Significant spatial differences in coupling coordination levels exist among different regions, with the highest level in the western regions, followed by the eastern regions, and the weakest in the central regions. This suggests an overall trend toward coordinated development between green finance and high-quality economic development in China, with green finance playing a significant role in promoting high-quality economic development. However, the growth rate of coupling coordination is relatively slow and exhibits regional heterogeneity. Lastly, drawing from these research findings, we put forward policy recommendations aimed at effectively advancing the development of green finance and high-quality economic growth in China.

With the development of internet technology, more two-sided market platforms, e.g., Tabao, Amazon and Lending Club, have emerged, and it is worth exploring the role that these two-sided market platforms can play in better serving users. This paper explores the impacts of platforms on participants’ trading strategies in the two-sided market, taking the online loan market as an example. Based on the management strategies of online lending platforms, we divided them into two categories: active online lending platforms and inactive online lending platforms. Evolutionary game theory and numerical simulation were employed to investigate the influences of these two types of platforms on the trading strategies of lenders and borrowers. We found that active and appropriate management measures taken by active online lending platforms can steer lenders and borrowers towards adopting win-win trading strategies, thus attracting more high-quality borrowers and corresponding investors. These results imply that a platform with two-sided market characteristics plays a crucial role in coordinating and guiding participants on both sides, and active management measures adopted by the platform can promote the benign operation of the two-sided market.

In knowledge distillation, a student model is trained with supervisions from both knowledge from a teacher and observations drawn from a training data distribution. Knowledge of a teacher is considered a subject that holds inter-class relations which send a meaningful supervision to a student; hence, much effort has been put to find such knowledge to be distilled. In this paper, we explore a question that has been given little attention: "when to distill such knowledge." The question is answered in our work with the concept of model calibration; we view a teacher model not only as a source of knowledge but also as a gauge to detect miscalibration of a student. This simple and yet novel view leads to a hard gate knowledge distillation scheme that switches between learning from a teacher model and training data. We verify the gating mechanism in the context of natural language generation at both the token-level and the sentence-level. Empirical comparisons with strong baselines show that hard gate knowledge distillation not only improves model generalization, but also significantly lowers model calibration error.

Two-sided markets serve as information intermediaries by connecting participants on both sides. In this study, we focus on the coordination of participants in the P2P lending market using a coupon strategy as an incentive to attract investment. Using a two-sided market model, we find that when a platform adopts the coupon strategy, (i) the platform utility and participants utility are both greater and (ii) the number of participants is greater. In addition, as most research on two-sided markets and coupon strategy focuses on theoretical models, our study provides empirical support using data from Renrendai.com over 2018 to 2019.

By data of 10,357 individual investors from Renrendai.com, an established peer‐to‐peer (P2P) lending platform in China over the period of 2011–2014, we examine whether investors’ prior investment outcomes influence their subsequent risk‐taking behavior in the credit market. We find strong evidence that a prior trading loss induces greater subsequent risk‐taking. Specifically, investors decrease their number of bids and the bidding amount; and choose listings with a lower interest rate and higher credit grade. The investors who obtain more prior gains become more cautious and take on less posterior risk. Overall, this study complements the lack of relevant research in the credit market.