Amid intensifying ecological crises and evolving legal frameworks for the green transition, intellectual property protection (IPP) has emerged as an important institutional guarantee for promoting corporate green innovation and environmental responsibility. However, stronger green innovation incentives and market attention may encourage firms to pursue strategic patenting for institutional benefits, creating green innovation bubbles (GIBs). Using the National Intellectual Property Demonstration City Policy as a quasinatural experiment, this study employs a multiperiod DID model to examine the impact of IPP on corporate GIBs. Results reveal that IPP significantly reduces GIBs. Mechanism analysis demonstrates that this effect operates easing financing constraints. Heterogeneity analysis indicates that the effect is greater among firms with higher green media attention, legally experienced CEOs, and nonheavily polluting industries. This study has implications for improving intellectual property governance and promoting high-quality green innovation.
This study examines whether frontier-based efficiency indicators, integrated with explainable machine learning, improve corporate financial distress prediction across contrasting institutional environments in China and the United States. Using 12,000 firm-year observations from 2015–2022, we augment standard accounting ratios with sector-specific Data Envelopment Analysis scores and Malmquist productivity indices in gradient-boosting classifiers. Incorporating efficiency measures increases out-of-sample AUC from 0.79 to 0.83 in China and from 0.82 to 0.84 in the US. Under the study’s staggered predictor design, prior-year DEA scores provide distress-relevant information in capital-intensive sectors before current-year ratios are finalized. SHAP explanations exhibit greater ranking stability than LIME under structural breaks, particularly around the COVID-19 shock, and rank technical efficiency second by mean absolute SHAP value. These findings suggest that DEA-based signals can provide interpretable early-warning information for credit analysts, portfolio managers, and regulators, with greater incremental value in lower-transparency markets.
With the start of the digital economy, standardised development of data commerce has shown considerable advantages for local financial innovation. Panel data from the 31 provinces, autonomous regions and municipalities of China (excluding Hong Kong, Macao and Taiwan) from 2015 to 2024 are used in this paper for a panel regression analysis to investigate the impact of changes in data trade regulations on the level of regional financial innovation efficiency and to study the moderating effect of different regional data resource availability and digital governance capacity. Based on the above results, it can be seen that data transaction supervision will help boost the productivity of regional financial innovation and provide stable support for the development of financial innovation projects through explicit rules and factor guidance. The amount of data resources and the level of digital governance are both positively correlated with this link, and they work together. Good Digital Governance can make more effective use of the power of data assets and enhance the good effects of data transaction supervision. Heterogeneity: Due to data-transaction regulations in the East, the increase in the level of financial innovation has been more pronounced. Due to the initial constraints, there will be no joint regulations for the central and western areas. Based on the above research results, some support has been provided for the optimisation of the data-transaction supervision system and high-quality development of digital finance and all-round construction of regional financial systems has been promoted.
Corporate environmental investment captures firms’ realized use of resources for environmental projects rather than the source of financing. Using 4,139 firm-year observations from 681 Chinese A-share listed firms during 2016–2024, this study examines the association between corporate environmental investment intensity (CEII) and Huazheng ESG ratings using firm fixed-effects models. CEII is positively associated with overall ESG ratings and each E, S, and G pillar. Supplementary regressions show that CEII is associated with lower normalized borrowing costs and greater R&D investment; borrowing costs are negatively associated with ESG ratings conditional on CEII, whereas the conditional R&D coefficient is positive but insignificant. The association remains positive after pooled 1%–99% winsorization and industry-by-year fixed effects. On a common 3,297-observation sample, lagged CEII remains positively associated with subsequent ESG ratings, but with smaller coefficients than the contemporaneous specification. All results are interpreted as conditional associations rather than causal effects.
How does the financial engineering behind China’s land-driven urbanization shape the urban–rural welfare gap? This study conceptualizes local governments’ borrowing against land assets as a distinct fiscal posture, proxied by the ratio of local financing vehicle debt to land conveyance income, and estimates its effect on within-prefecture urban–rural inequality across Chinese cities. Fixed effects and instrumental variable estimates reveal a consistent trajectory wherein the more heavily a jurisdiction leverages its land, the wider its urban–rural income divide becomes, an association that remains robust to a range of specification tests. Two mechanisms explain this result. First, rather than seeding the rural cooperatives that could channel urban growth back to the countryside, land-based borrowing crowds them out, weakening the organizational capital on which rural incomes depend. Second, the rural roads it finances do not function as conduits of shared prosperity but as channels drawing labor and capital toward cities. Quantile estimates reveal a nonlinearity whereby the widening force is strongest where the gap remains modest, then weakens and eventually reverses once the divide grows large.
We test whether mutual fund managers intentionally exploit capital market anomalies by decomposing changes in fund factor loadings into active trading and mechanical drift from changes in the factor loadings of held stocks. Funds sorted into the highest loading decile earn significant risk-adjusted outperformance relative to the lowest. Yet funds do not trade in the direction of any of the seven anomaly factors studied, whether classified by factor loadings or underlying firm characteristics. Changes in stock-level factor loadings explain 60%–86% of the variation in fund factor loadings, while trading explains virtually none. Non-traded holdings drift significantly between quarters, confirming the mechanical channel by construction. Anomaly exposure in mutual funds reflects market-wide stock dynamics rather than deliberate strategy.
Precautionary savings theory predicts that rising external uncertainty leads firms to hold more cash. We show that a different pattern emerges under a persistent structural shock. U.S.–China technology decoupling significantly lowers corporate cash holdings. We construct a firm-level measure of decoupling exposure from cross-border patent citations and test it on Chinese A-share listed firms over 2007–2023. A decomposition shows that the decline reflects neither falling revenue nor higher dividends, and that neither capital expenditure nor the net change in cash moves significantly. Rather, the evidence is consistent with a compositional reallocation of liquidity toward working capital: more exposed firms hold more inventory and receivables and have longer working-capital cycles, together with a more conservative financing posture, so the cash share falls. The precautionary motive does not disappear; its carrier shifts from financial to operational liquidity. The pattern is strongest among non-state firms, technology-intensive firms, and firms with concentrated supply chains. Decoupling lowers innovation quantity in the short run but shifts innovation composition toward higher quality.
This study examines how artificial intelligence (AI) credit scoring shapes financial inclusion for Small and Medium Enterprises (SMEs) in China and Kyrgyzstan, two economies with sharply divergent digital ecosystems. A novel panel dataset of 850 SMEs tracked from January 2022 to December 2026 is analyzed using six machine learning architectures: Support Vector Machine (SVM), Random Forest, XGBoost, Convolutional Neural Network (CNN), Long Short-Term Memory (LSTM), and Transformer. Employing 35 alternative-data variables sourced and harmonized across both regulatory environments, each model is evaluated on classification accuracy, Root Mean Square Error (RMSE), Area Under the Curve (AUC), F1-score, and false rejection rate. The Transformer achieves the highest classification accuracy (94.5%) and lowest RMSE (0.015) in China's data-rich ecosystem, yet suffers a 22.2-percentage-point degradation in Kyrgyzstan, where a 31.5% average missing-data rate raises false rejections to 28.5%. XGBoost proves the most resilient architecture in the data-sparse context. Drawing on Institutional Theory, this study demonstrates that algorithmic credit-scoring outcomes are contingent on formal digital infrastructure, regulatory quality, and informal economic norms. A five-phase Strategic Policy Framework for Equitable Algorithmic Finance is proposed, accompanied by a Localized Technology Assessment Framework, to guide regulators and FinTech developers in transitional markets.
