Chinese convertible bonds are examined in two settings: portfolio construction and relative-value trading. At the portfolio level, a convertible bond index is compared with the HS300 equity index and a broad bond index over 2005–2024 using return, volatility, Sharpe ratio, maximum drawdown, illustrative allocations, and mean–variance frontiers. The convertible bond index earns a return close to equities with lower volatility and a smaller drawdown, while allocations containing convertibles record higher Sharpe ratios than the stock–bond benchmark. At the trading level, mispricing is defined as the gap between GARCH volatility estimated from the underlying stock and the implied volatility embedded in the bond price. A positive gap indicates that the conversion option is priced at a lower volatility than the stock estimate. Following a static delta-hedged Feikai CB example, fixed market-wide thresholds are compared with bond-specific expanding-window percentile rules based only on information available at each date. With a P50 exit, clean-convergence trades increase from 945 to 1,415 and the timeout share falls from 25.0% to 9.4%. The contrast between CITIC and Wencan shows that spread dispersion, rather than spread level alone, is important for determining whether apparent mispricing can be traded.

This paper asks whether the structural consequences of High-Speed Rail (HSR) expansion arise from a city's first connection to the network or from subsequent network densification. This paper matches a hand-coded database of passenger-rail line openings to a panel of Chinese prefecture-level cities from 2008 to 2019 and exploits within-city changes in access in a two-way fixed effects framework. First HSR access increases the tertiary-sector share of Gross Domestic Product (GDP) by 0.925 percentage points and reduces the secondary-sector share by 1.078 percentage points. The reallocation is most pronounced in western cities. By contrast, once initial access is accounted for, additional lines and hub status yield statistically imprecise average effects. A pretreatment placebo and a 2014-cohort difference-in-differences design corroborate the timing and direction of the baseline estimates. These findings identify network entry, rather than subsequent densification, as the principal margin through which HSR expansion reshapes urban industrial structure. They imply that the structural returns to extending network coverage can differ materially from those of adding capacity and connectivity within already-served cities.

Cross-border live streaming e-commerce has emerged as a rapidly growing channel for international trade, yet the role of cultural differences in shaping consumer purchase decisions within this context has received limited empirical attention. This study investigates how cultural differences influence consumer purchase decisions in cross-border live streaming environments, grounded in Hofstede's cultural dimensions theory and the stimulus-organism-response framework. A conceptual model is developed to examine the mechanisms through which cultural dimensions—including individualism-collectivism, uncertainty avoidance, and power distance—affect consumer trust in streamers, perceived product authenticity, and, ultimately, purchase decisions. The analysis reveals that cultural differences moderate the effectiveness of streamer communication styles, the formation of consumer trust, and the weight of social influence in purchase decisions. Specifically, consumers from collectivist cultures respond more strongly to streamer-consumer interaction and community cues, while those from high uncertainty avoidance cultures place greater emphasis on product certification and detailed demonstrations. The findings contribute theoretically to the cross-cultural e-commerce literature and offer practical implications for cross-border live streaming practitioners seeking to tailor their strategies to diverse cultural markets.

Traditional comprehensive budget management suffers from multiple shortcomings such as the absence of strategic orientation, budget slack, and insufficient organizational coordination, making it difficult to adapt to complex and dynamic business environments. Digital technologies such as artificial intelligence and big data provide a technological path for the transformation of budget management. This paper systematically reviews the mechanisms and application scenarios of AI technologies-including time-series forecasting, anomaly detection, causal inference, reinforcement learning, and large language models-throughout the entire process of budget preparation, execution monitoring, analysis and adjustment, and assessment and review. The research shows that AI, relying on data-driven approaches, can achieve accurate forecasting, real-time risk identification, quantitative attribution, and dynamic resource optimization, effectively resolving the inherent pain points of traditional budget management. From the four dimensions of data, technology, organizational talent, and institutional risk control, this paper constructs a guarantee system for the implementation of AI budgeting, while also pointing out the practical obstacles currently faced by intelligent budget transformation, such as insufficient model interpretability, data silos, a shortage of compound financial talent, and imperfect supporting institutions. This paper enriches the theoretical system of intelligent budget management and provides implementation insights for enterprises to advance the digital-intelligent transformation of budgeting.

Agency conflict arising from information asymmetry remains a vital source of governance conflict in corporations, since the insiders usually have access to more relevant and up-to-date information than the external investors, creditors, regulators, and minority stakeholders. In this work, this study presents a new artificial intelligence-based governance intelligence solution that is capable of uncovering information asymmetry gaps, constructing governance relationship networks, and estimating the level of information transparency in each firm-year period. With Chinese A-shares firms from the time interval 2018 to 2024 as the research sample, our model draws its features from annual reports, internal control reports, board statements, regulatory inquiries, ownership structure, management compensation, analyst forecast, and stock trading. As for the methodology, the model utilizes DeBERTa-v3 to extract the governance disclosure information, Sentence-BERT to find out any inconsistency among disclosure variables, Neo4j and graph attention networks to construct the relationship-risk embedding, and LightGBM to classify firms with a higher risk of information asymmetry. From our empirical experiments, it can be concluded that the whole AI-based model yields AUC at 0.874 ± 0.018 and macro-F1 at 0.812 ± 0.021, which is an improvement of 0.116 compared to the conventional governance metrics in terms of AUC.